OpenAI picked an awkward week to celebrate.

On August 31 the company announced that ChatGPT Ads had reached USD 1 billion in annualised revenue run rate, roughly 200 days after launch, and opened self-serve buying to advertisers across India, Europe, the Middle East and North Africa.

Ads now run in more than 40 countries, served to free-tier and Go subscribers who make up the bulk of ChatGPT’s roughly one billion weekly users.

It is the sort of number a company publishes when it wants prospective investors to see a business with more than one engine. OpenAI has filed confidentially for a US listing and carries a valuation reported at around USD 852 billion.

Analysts were quick to note the milestone also confirms a miss. OpenAI told investors in April it expected around USD 2.5 billion in advertising revenue this year. A billion-dollar run rate in late August does not get it there.

That, however, was not what dominated the coverage. Two other developments landed in the same 24 hours, and both point at problems that no amount of ad inventory will fix.

Apple’s forensic turn
In a filing in the US District Court for the Northern District of California, Apple alleged that Chang Liu, a former senior system electrical engineer at the iPhone maker and a named defendant in its trade secret case, accessed a power converter circuit schematic while working at OpenAI, and trained an artificial intelligence agent on proprietary Apple material in March 2026.

The sting is in the provenance. Apple said the evidence surfaced on a MacBook that OpenAI itself handed over on 21 August as part of the discovery process. Having examined the device, Apple has asked the court to expedite discovery, which is rarely the move of a plaintiff who thinks its case is thin.

Apple sued on July 10, naming OpenAI, its hardware subsidiary io Products, chief hardware officer Tang Yew Tan and Liu.

The complaint alleges Liu retained an Apple-issued laptop after leaving in January 2026 and exploited an authentication weakness to pull engineering documents and specifications for unreleased products while already on OpenAI’s payroll.

Apple further claims OpenAI coached departing staff on how to sidestep exit security procedures, and that it asked hardware partners to apply a metal finishing technique Apple developed while leaving those partners with the impression that permission had been granted.

OpenAI filed a 31-page motion to dismiss on August 5, calling the complaint baseless. A hearing on Apple’s request for a preliminary injunction is listed for October in San Jose. No court has ruled on the merits of anything.

The commercial relationship survives, uncomfortably.

ChatGPT remains an option inside Siri and Apple Intelligence, and Apple’s complaint expressly carves the integration agreement out of the dispute. Two companies are litigating over the hardware category that could eventually make that partnership irrelevant, while continuing to earn from it.

The backdrop is OpenAI’s roughly USD 6.5 billion purchase of io Products in 2025, which put Jony Ive and a cohort of former Apple engineers inside the company that Apple now accuses of systematic misappropriation.

The Cursor cutoff
The second development is smaller in dollar terms and considerably larger in what it signals.

On August 28, OpenAI notified SpaceX that it would wind down the contract supplying its models to Cursor, the AI coding editor, with a proposed shutoff date of November 12. It will supply no new models during the wind-down.

Cursor’s parent, Anysphere, was among OpenAI’s earliest customers, and the two worked together for close to four years. What changed was the ownership column.

SpaceX announced a USD 60 billion all-stock acquisition of Anysphere on June 16 and completed it on August 14, converting Anysphere stock into roughly 389 million SpaceX Class A shares.

Cursor, which had climbed from about USD 100 million in annualised revenue in January 2025 to roughly USD 4 billion by mid-2026, now sits inside SpaceX’s AI division.

OpenAI’s stated reason was trust rather than performance. It said it could not be confident SpaceX would use its technology within its terms of service, citing a history of contract breaches at Musk-owned businesses.

The company pointed to Twitter’s (now X) conduct after Musk’s takeover and, more pointedly, to Musk’s own admission under oath earlier this year that xAI had violated those terms.

It also invoked its forthcoming model, Astra, which OpenAI said in August may reach the critical cyber capability threshold under its own preparedness framework, as grounds for tighter control over who receives access.

Its custom agreement with Cursor, it noted, allowed only a narrow window to cancel following a change of control, and it used the full notice period available.

Cursor co-founder Michael Truell, now a SpaceX executive, was conciliatory, saying OpenAI models account for roughly 5% of Cursor traffic and that discussions were under way.

Musk was not. On X he said he could not care less, described Altman and OpenAI president Greg Brockman as untrustworthy, and repeated his charge that the pair took an open source nonprofit.

Anthropic moved within hours, with co-founder Tom Brown signalling additional compute for Claude models inside Cursor.

A decade of grievance
To understand why a routine contract termination read as an act of war, you have to go back to 2015.

Musk and Altman founded OpenAI together that year as a nonprofit, explicitly to stop advanced AI ending up under the control of any single company. Musk contributed tens of millions of dollars. By 2017 the founders had concluded they could neither raise the capital nor recruit the researchers needed to compete without a commercial arm.

Musk accepted the diagnosis but wanted the cure on his own terms, proposing either a majority stake in the for-profit entity or folding the whole thing into Tesla.

Altman, Brockman and Ilya Sutskever refused. Musk left the board in 2018, took talent with him and stopped the donations he had promised.

The grievance stayed largely private for five years. It went public in July 2023 with the launch of xAI, and became litigious in February 2024, when Musk sued in California state court.

He withdrew that case in June and refiled in federal court in August, alleging that Altman and Brockman had manipulated him into bankrolling a charity they then converted for personal enrichment.

January 2025 added the personal insult. Altman appeared alongside Donald Trump to announce Stargate, a USD 500 billion AI infrastructure programme, without warning Musk, who was by then running the Department of Government Efficiency.

Weeks later a Musk-led consortium bid USD 97.4 billion for OpenAI’s assets. The board rejected it within hours. Altman replied on X with an offer to buy Twitter for USD 9.74 billion, a number chosen for the joke.

OpenAI countersued in April 2025, calling the bid a feint and describing a sustained campaign of harassment waged through press attacks, a pretextual records demand and posts to more than 200 million followers on a platform Musk owns.

A judge let those counterclaims proceed in August 2025, and they remain live.

Then the fronts multiplied. In August 2025, xAI and X sued Apple and OpenAI in Texas, arguing that the Siri partnership made it impossible for any rival assistant to reach the top of the App Store.

A month later xAI sued OpenAI over engineer poaching and trade secrets. That case was dismissed in February 2026 with leave to amend, dismissed again in June without it, and is now before the Ninth Circuit. The antitrust case survived dismissal in November 2025 and is still running.

The trial that settled nothing
The main event opened in Oakland on 28 April 2026, with Musk seeking as much as USD 134.5 billion from OpenAI and Microsoft.

Three weeks of testimony put Altman, Brockman, Satya Nadella and Musk himself on the stand, and produced one genuinely uncomfortable moment on  April 30, when Musk conceded under cross-examination that xAI had partly used OpenAI’s outputs to train its own models. Distillation, he argued, was standard industry practice.

On 18 May a nine-person advisory jury took under two hours to find that Musk had waited too long to sue.

Judge Yvonne Gonzalez Rogers adopted the verdict and dismissed the claims from the bench.

Nobody ruled on whether Altman and Brockman had actually broken a promise. Musk called it a calendar technicality and confirmed an appeal.

That outcome explains almost everything since. The verdict removed the last serious threat to OpenAI’s corporate structure ahead of a listing, but it left the accusation entirely intact and Musk free to keep repeating it.

By July the two were trading insults on X, Musk reviving his scammer nickname, Altman needling him over orbital data centres, Musk suggesting Altman might one day need a parole officer.

When Apple sued in July, Musk amplified it within days.

Leverage, not lawsuits
Musk’s position has changed materially in the interim. SpaceX absorbed xAI in February 2026 in a deal valuing the combined group at USD 1.25 trillion, listed on June 12 in the largest IPO on record, and moved on Cursor four days later.

The man suing OpenAI over its commercialisation now runs a public conglomerate spanning rockets, satellites, a social network, a frontier lab and the leading AI coding tool.

That is why the Cursor decision matters beyond its size. Model access has quietly become leverage. Anthropic cut OpenAI off from Claude in August 2025 and later restricted xAI.

OpenAI has now done the same to a Musk-owned business. Each move was defensible on its own terms. Together they establish that a frontier lab will treat supply as a strategic instrument, and that a change of control can rewrite a critical dependency without warning.

Cursor will probably absorb it. OpenAI serves a twentieth of its traffic, Anthropic is stepping in, and SpaceXAI has its own Grok models to fall back on.

The lasting effect will be felt in procurement offices, where single-vendor exposure in the AI stack is about to get a much harder look.

OpenAI, for its part, is holding three things together at once. It wants a clean narrative for public markets. It is defending a trade secret case whose discovery could force disclosure of its hardware roadmap.

And it is managing a rival who lost in court, is appealing, owns the venue where the argument plays out, and now controls a tool its own customers open every morning.

The advertising billion was meant to be the story. It lasted about a day.

The post OpenAI’s advertising billion arrives as Apple and SpaceX tighten the screws appeared first on International Finance.

OpenAI’s newest and most powerful AI system, called GPT-5.6, has been released only to a small group of hand-picked companies. And it is not entirely OpenAI’s choice. The Donald Trump administration asked OpenAI to limit the rollout to a small set of United States government-approved partners before any wider release, citing security concerns, marking the first time the US pre-emptively asking an American AI company to restrict a model launch.

The decision came from the very top of the administration. The White House’s Office of the National Cyber Director and Office of Science and Technology Policy made the request as the administration works to build a framework for testing and evaluating the security of new models.

Commerce Secretary Howard Lutnick was also closely involved, reportedly wanting to ensure that all relevant sections of the government had tested and cleared the model before it went live.

Why is the government so concerned?
GPT-5.6 is not a routine upgrade. To understand why Washington has intervened so forcefully, it helps to understand what the model can actually do.

Under OpenAI’s own internal safety framework, all three versions of GPT-5.6 are classified as ‘High’ capability in both cybersecurity, and biological and chemical risk. In plain terms, the model is considered powerful enough that, in the wrong hands, it could potentially be used to launch sophisticated cyberattacks, probe critical infrastructure, or assist in the development of dangerous biological or chemical agents.

The biology numbers alone are striking. GPT-5.6 scored 53.5% on the Virology Capabilities Test, 60% on Molecular Biology, 68.4% on Human Pathogen Capabilities, and 68.3% on a benchmark called World-Class Bio, approximately nine percentage points above its predecessor, GPT-5.5.

These are not abstract academic scores. They represent a measurable increase in the model’s ability to reason about biological systems in ways that security experts consider ‘sensitive’.

On the cybersecurity side, GPT-5.6 Sol shifts the performance frontier for long-horizon security tasks, including vulnerability research and exploitation. On ExploitBench, a standard industry test, it performs on a par with Anthropic’s most advanced model while using roughly a third of the output tokens.

OpenAI’s own pre-release testing also found what are described as universal jailbreaks – systemic attack vectors capable of bypassing safeguards across the model family – before any public access was granted. That finding alone would give any government cause for concern.

An independent pre-deployment evaluation by METR found that Sol’s detected reward-hacking rate, a form of AI behaviour where the model games its own tests rather than genuinely solving them, was the highest of any public model METR has tested.

Put together, these findings paint a picture of a system that is extraordinarily capable but also, by its creators’ own admission, not yet fully understood.

What exactly is GPT-5.6?
OpenAI released three versions of the model named Sol, Terra, and Luna. Sol is the most powerful of the three. Terra offers a balance of efficiency and power, while Luna is designed for speed and affordability.

According to OpenAI, Terra is intended to offer performance comparable to its predecessor GPT-5.5 while costing half as much, making it potentially the most commercially attractive option for businesses once the restrictions are lifted.

Sol, however, is where the real leap forward lies. It introduces a ‘max’ reasoning effort mode and an ‘ultra’ mode that uses coordinated sub-agents to solve highly complex tasks.

Sub-agent orchestration, in everyday language, means the AI can break a complex task into parts and assign each part to a specialised mini-version of itself working in parallel, a significant advance in how much autonomous work the system can handle without human involvement.

In benchmark testing, Sol set a new state-of-the-art score on Terminal-Bench 2.1, which tests command-line workflows requiring planning, iteration, and tool coordination, achieving 88.8%, rising to 91.9% in its ultra configuration, ahead of competing frontier models.

The current limited preview is available to approximately 20 companies, whose participation has been approved by the United States government.

OpenAI’s uncomfortable position
OpenAI has complied with the restrictions but has made its displeasure clear. CEO Sam Altman told staff in a memo that the Trump government is approving access ‘customer by customer’,” and he hopes to release GPT-5.6 more broadly a couple of weeks later.

The company’s public statement was pointed.

“We don’t believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them,” OpenAI said.

OpenAI was careful, however, not to frame the situation as a conflict with the White House. The company said it believes the Trump administration still has the best interests of US AI competitiveness in mind, and positioned the restricted launch as a short-term concession during an in-between period, one where the government has announced a plan to evaluate new model releases, but has yet to detail exactly how that process will work.

The company’s preferred framing is that it is helping build the scaffolding, not being caught in it. OpenAI described the restricted rollout as a short-term step, and said it represents the strongest path to broader availability in the coming weeks, as the company works with the administration to develop a cybersecurity executive order framework, and a repeatable process for future model releases.

Now, a report from British daily Financial Times (FT) claims that OpenAI has discussed giving the Trump administration a 5% stake. The Altman-led firm also wants other American AI companies to hand over similar stakes to the government.

In fact, Trump, in June 2026, said that he was exploring options to give the public a stake in leading AI companies in response to concerns that individual Americans will not share in the sector’s expected profits.

If Altman’s proposal gets serious consideration at the White House, the result will be the creation of an investment vehicle similar to the “Alaska Permanent Fund,” a state-owned corporation seeded with oil revenues that pays annual dividends to residents and helps support Alaska’s budget.

While OpenAI earlier proposed the creation of a “public wealth fund” to invest in AI companies and distribute proceeds to citizens, Anthropic has been actively exploring a “digital dividend,” defined as payments to Americans funded by taxes on the AI sector.

Altman has ⁠reportedly discussed the stake sale with Trump, Commerce Secretary Howard Lutnick, and Treasury Secretary Scott Bessent, apart from engaging with Democratic Senator Bernie Sanders in recent weeks.

A pattern taking shape across the industry
This is not an isolated incident, and OpenAI is not the only company navigating this new terrain.

The Trump administration previously placed an export control order on Anthropic, leading the AI company to pull its latest advanced models, Mythos and Fable, from public access after concerns arose in Washington and on Wall Street over their advanced cybersecurity capabilities. Anthropic’s Mythos model has since been made available only to a curated group of trusted organisations through a programme called Project Glasswing.

After Anthropic released its most powerful public model, “Fable 5,” the administration ordered the company to remove access for any foreign national, prompting Anthropic to take the model down entirely, raising questions about how much power the government should have over AI model releases.

The legal basis underpinning all of this is relatively recent. A Trump executive order signed on 2nd June 2, 2026 requires federal benchmarking of new AI models, and is the formal mechanism through which the government can gate these rollouts.

The order technically asks companies to voluntarily submit advanced models for government review up to 30 days before release, but critics say the voluntary label is increasingly difficult to distinguish from a mandatory one.

Dean Ball, a former White House AI adviser, argues that the executive order has effectively created a de facto involuntary licencsing regime for frontier AI, and that without clearly defined safety standards, endless launch delays could hand China an advantage in the global AI race while also jeopardising the billions of dollars flowing into AI infrastructure.

What comes next
The immediate commercial stakes are significant. Enterprises that had been anticipating GPT-5.6 for deployment in cybersecurity defence, drug discovery, or large-scale software engineering will need to wait.

During the preview period, OpenAI said it will work with enterprise customers on longer-term approaches, including the possibility of adding privacy-preserving detection, customer-operated safety controls, and calibrated access.

For the broader AI industry, the implications are harder to quantify but potentially more consequential. If Washington’s current approach becomes the template, with government review before commercial release, access approved company by company, and restrictions enforced through executive order rather than legislation, the relationship between frontier AI labs and the state will look fundamentally different within a year.

OpenAI has framed the restricted launch as a temporary inconvenience on the road to a broader rollout. Whether Washington sees it the same way remains to be seen. What is clear is that the era in which AI companies could release transformative systems entirely on their own terms, on their own timetables, may already be over.

The post White House’s tech ‘lock and key’ strategy shifts to OpenAI appeared first on International Finance.

World hit the headlines in 2023, when they introduced the World ID app as a dedicated proof-of-human concept across the internet. World ID app, as a standalone application, also represents a shift from the company’s previous wallet-integrated approach to identity confirmation.

World’s flagship product, the iris-scanning Orb device, scans and helps verify humans for use across an array of online applications. World incentivised use of the Orb with its Worldcoin (WLD) crypto token. However, it further expanded its proof-of-human suite to include other verification methods.

Apart from announcing partnerships with Tinder and Zoom, World has also launched “Concert Kit,” a tool powered by World ID that enables artists to reserve tickets for verified humans. The goal is clear: fighting bot-driven ticket scalping by making human verification mandatory for event access.

As part of the partnership, Match Group, Tinder’s parent company, will expand its existing World ID tie-up to serve American users. Zoom, the video meeting app, will integrate World’s deepfake detection technology to spot fakes; fund manager VanEck and a few other firms are already testing the technology. San Francisco-based DocuSign will offer World ID support to ensure digital signatures get done by actual persons only.

Stating that Worldcoin’s network has reached 18 million verified humans across 160 countries, Tom Lee, Chairman of Ethereum treasury firm BitMine Immersion Technologies and board member of Worldcoin treasury firm Eightco, said, “Proof of human and verified human identity vaulted to a critical priority for social networks and banking and financial systems as AI and agentic-AI capabilities experienced an exponential step forward in the past few months.”

World is also introducing tiered verification, with Orb scans at the top, a government ID scan in the middle, and a new selfie-based option at the bottom. World will further expand Orb’s availability across New York, Los Angeles, and San Francisco.

The post Sam Altman-backed World ties up with Zoom, Tinder appeared first on International Finance.

Aguilar, who leads the Los Angeles–based nonprofit LatinoProsperity, feared the move would betray OpenAI’s founding mission to “benefit all of humanity” by empowering investors to reap unlimited private profits.

That moment spurred Aguilar to action. He began speed-dialling allies in California’s philanthropic and civil rights circles, determined to scrutinise OpenAI’s restructuring plan and its implications for the public interest.

Over the ensuing months, a broad coalition of more than 50 advocacy organisations coalesced: labour unions, community foundations, tech accountability groups. All rallied around a shared concern that OpenAI’s transformation from a nonprofit lab into a profit-driven corporate behemoth could set a dangerous precedent.

By January 2025, they were urging California’s attorney general to intervene, warning that billions in charitable assets (the ones intended for the public good) were at risk of being effectively privatised.

What had begun as a little-noticed corporate restructuring proposal was quickly snowballing into a high-stakes battle over the future of OpenAI, the governance of artificial intelligence, and the integrity of the nonprofit system itself.

From idealistic lab to tech titan

OpenAI’s origin story is steeped in idealism. The San Francisco research lab launched in late 2015 as a nonprofit venture backed by tech luminaries including Sam Altman and Elon Musk, who together pledged $1 billion to fund it. At the time, concerns were growing that AI development was dominated by a few big tech firms driven by profit.

OpenAI’s founders vowed a different path: build advanced AI in the service of all humanity and share the research openly. Its charter emphasised long-term social benefits over financial gain, even declaring that when conflict arises, the nonprofit mission would “take precedence” over any obligation to generate profit.

For several years, OpenAI operated like an AI think tank, publishing cutting-edge research and freely sharing its code. But as the race to develop powerful AI accelerated, the organisation faced a dilemma. Training world-class AI models required far more computing power (and money) than its initial philanthropic funding could support.

By 2019, OpenAI’s leadership made a controversial pivot. They set up a hybrid structure by creating a for-profit arm, OpenAI Global, under the umbrella of the original nonprofit, OpenAI. This allowed them to attract venture capital while ostensibly keeping the nonprofit’s oversight and mission intact.

This compromise introduced a novel “capped-profit” model. Investors could earn returns on their money, but those returns were capped at a certain multiple. After that point, the nonprofit would retain surplus value to fund its mission. The idea was to access billions in Silicon Valley capital without fully sacrificing OpenAI’s altruistic DNA.

In practice, the hybrid model allowed OpenAI to tap deep-pocketed backers. Microsoft alone poured in a reported $13 billion (across multiple funding rounds) for a share of OpenAI’s technology and profits.

By late 2023, OpenAI’s ChatGPT had become a global sensation, drawing over half a billion weekly users and solidifying OpenAI as a leading AI provider.

Yet the influx of capital came with mounting pressure to compete and monetise. OpenAI began behaving more like a Silicon Valley startup than a pure research outfit. It charged for API access, launched paid subscriptions, and curtailed its once-open research disclosures for “security and competitive” reasons. The tension between its nonprofit ideals and market ambitions grew harder to ignore.

The breaking point came in November 2023 with an extraordinary boardroom drama. OpenAI’s nonprofit board, tasked with ensuring the company stayed true to its mission, abruptly fired CEO Sam Altman, citing a “lack of consistent candour” in his communications.

The shock ouster of Altman, the charismatic figurehead of ChatGPT’s rise, sent the tech world into a tailspin, especially after it emerged that concerns about AI safety and OpenAI’s rapid pace might have been at issue.

Within five days, Altman was reinstated following an employee and investor uproar. Most of the board members who had ousted him resigned under pressure. The episode was a stark illustration of OpenAI’s governance quandary. The nonprofit oversight that reassured the public about OpenAI’s benevolent mission had become a wildcard factor in a company now valued like a $100 billion startup.

As Vox observed, the saga “made it clear that the nonprofit’s control of the for-profit could potentially have huge implications,” particularly for Microsoft, which had billions at stake. After that turmoil, OpenAI’s leadership grew even more convinced that its existing structure was unsustainable.

Shift to a public-benefit corporation

In late 2023, OpenAI started mapping out a plan to overhaul its corporate structure. The centrepiece would be converting OpenAI’s for-profit subsidiary into a new entity incorporated as a public benefit corporation (PBC). A PBC is a company that balances profit-making with a stated social mission.

Crucially, unlike the earlier capped-profit model, this change would remove the hard ceiling on investor returns, allowing venture backers to profit without limit. OpenAI’s nonprofit parent would likely either become a minority shareholder in the new PBC or receive a one-time payout. In other words, OpenAI was preparing to shed the very safeguards that once made it unique: the nonprofit’s veto power and the cap on profits, all in favour of a more typical corporate arrangement.

Developing advanced AI has become a capital-intensive arms race, with rivals like Google, Meta, Anthropic, and Elon Musk’s new startup xAI all vying for talent and computing resources. OpenAI’s leadership argues that staying ahead in AI requires access to far greater funding than a nonprofit model can provide. Under American law, nonprofits face strict limits on raising investment.

Neil Elan, a partner at a tech law firm, stated that “they can’t sell stock or offer returns.”

He explains that “equity is what drives a lot of these high-valuation models in Silicon Valley. Without the ability to issue lucrative shares, OpenAI feared it wouldn’t fully compete with Meta, Microsoft, and Google, which have access to a lot more resources… without comparable funding.”

Those worries crystallised as OpenAI negotiated a blockbuster financing deal earlier this year. In April 2025, OpenAI announced it had closed a record-setting $40 billion funding round led by Japan’s SoftBank. The deal pegged OpenAI’s valuation at an eye-popping $300 billion. The catch? Roughly 75% of that investment is contingent on OpenAI completing its structural revamp by the end of 2025.

The investment agreement allows SoftBank and other backers to withdraw up to $30 billion of the funding if OpenAI does not transition to the new PBC structure on schedule. The message was clear: to secure the full war chest needed for its aggressive AI roadmap, OpenAI had to cast off any structural quirks that made investors nervous.

And the investors were getting nervous. The November boardroom fracas had highlighted how OpenAI’s nonprofit oversight could unpredictably intervene in business decisions. This created a risk factor almost unheard of among tech unicorns.

According to a previously unreported letter from OpenAI’s lawyers to California regulators, “many potential investors in OpenAI’s recent funding rounds declined to invest due to its unusual governance structure.”

This contradicts earlier narratives that investors were lining up in droves.

Indeed, some of Silicon Valley’s biggest players baulked at OpenAI’s hybrid model after seeing Altman briefly dethroned by a nonprofit board. To calm the investor concerns, OpenAI’s leadership initiated plans soon after the Altman episode to remove nonprofit control and restructure as a profit-centric PBC. In effect, the startup’s meteoric success was forcing it to become more like a traditional corporation in order to keep raising capital at sky-high valuations.

Sam Altman himself has acknowledged that, in hindsight, OpenAI’s founders underestimated how costly and fast-moving the AI race would become.

He noted that if he could redo 2015, he might have structured OpenAI differently from the start. Other AI startups, Anthropic (founded by ex-OpenAI researchers) and xAI (founded by Musk), learnt from OpenAI’s example and launched as public benefit corporations from day one.

Now OpenAI is racing to catch up and give its backers the standard corporate framework and eventual stock market payday they expect. Investors are already eyeing a potential OpenAI IPO by 2027, which could turn early stakes by firms like Microsoft and SoftBank into massive profits.

The activist backlash

As OpenAI moved forward with its restructuring behind closed doors, Orson Aguilar and his allies mobilised a counteroffensive in public. Aguilar’s first call after digesting the news in October 2024 was to Fred Blackwell, CEO of the San Francisco Foundation.

Blackwell, a prominent figure in Bay Area philanthropy, immediately recognised echoes of a past fight. In the 1990s, he and other advocates had taken on a wave of nonprofit hospitals and insurers attempting to convert into for-profit companies.

As those healthcare nonprofits demutualised, some executives manoeuvred to boost payouts for themselves while hollowing out the charitable foundations that were meant to receive the nonprofits’ assets.

Consumer advocates in California, including Blackwell’s colleague Judith Bell, latched onto an obscure but powerful statute. Under state law, all the assets of a nonprofit are irrevocably dedicated to the public and belong to the people of the state forever, only to be used for charitable purposes.

The law grants California’s attorney general broad authority to approve or deny any conversion of a nonprofit’s assets and to ensure that the public is protected when a nonprofit becomes a private entity.

Bell and others built a coalition in the 1990s that pressured the California Attorney General to enforce those rules rigorously. The result was a series of agreements that preserved enormous charitable endowments even as hospitals transitioned into for-profit status.

In California alone, that activism helped create three of the state’s largest private foundations, including the $4 billion California Endowment, which was funded by assets spun out of former nonprofit hospitals.

Advocates estimate that they saved $15 billion in charitable funds from being diverted by businesses and their investors through those deals.

“You can protect the charitable assets and allow these companies to go forth in the for-profit world,” Bell says, reflecting on that chapter. In other words, compromise is possible. Companies can convert, but the public must get its fair share.

Seeing history repeat itself, Bell, Blackwell, and Aguilar decided to revive that old playbook. By late 2023, they had assembled a broad coalition of community and labour organisations. More than 50 groups, ranging from tech accountability nonprofits to unions like SEIU, expressed alarm at the idea of OpenAI’s charitable assets being diverted to private gain.

In January 2025, the coalition launched a public campaign and formally petitioned California Attorney General Rob Bonta to scrutinise OpenAI’s plans. Their request was direct: Do not approve this conversion without firm guarantees that OpenAI’s nonprofit assets (both tangible and intangible) will be fully valued and used for the public good going forward.

The coalition’s letter to Bonta, costeered by Aguilar, expressed doubt that OpenAI intended to comply with the spirit of the law. They accused OpenAI of skirting transparency and failing to detail how its nonprofit stake, which some estimate could be worth $20 to $30 billion given OpenAI’s valuation, would be protected.

Any scheme that values OpenAI’s nonprofit share at even a penny less than fair market value “would be unlawful,” Aguilar argues.

He adds that anything short of full independence for the nonprofit arm risks allowing commercial imperatives to override the charity’s purpose. In short, the activists want OpenAI’s nonprofit to remain firmly in control or receive a payout that reflects its foundational role in creating ChatGPT and other breakthroughs.

Ideally, they envision that endowment fuelling what could become one of the best-resourced nonprofits in history, a massive charitable foundation to fund AI for good, free from the influence of OpenAI’s new for-profit owners.

This grassroots pressure has already scored some symbolic wins. By March, as media coverage of the “OpenAI rebellion” intensified, the company’s leadership reached out to engage with the coalition. Aguilar, Blackwell, and others sat with OpenAI representatives in San Francisco for a tense meeting.

According to participants, OpenAI staff were eager to correct what they described as “misconceptions” about the restructuring and even asked for feedback on how the nonprofit’s mission might evolve in the future. But when Aguilar pressed for concrete answers on how much funding and independence the nonprofit would retain under the new plan, he says OpenAI’s emissaries deflected. Not long after, OpenAI announced it was creating a special advisory commission, a move widely seen as a response to the coalition’s campaign.

California’s attorney general enters the fray

California Attorney General Rob Bonta, who oversees nonprofit charities in the state, became a pivotal figure in this unfolding drama. Triggered by the January petition, Bonta’s office quietly opened an investigation into OpenAI’s plans and requested financial records from the company earlier this year.

While such probes are typically confidential, Bonta’s spokesperson publicly confirmed in the spring that the California Department of Justice was actively reviewing OpenAI’s restructuring and remained in continued conversations with the company.

OpenAI must convince Bonta that its conversion plan will not improperly discard any charitable trust obligations. The attorney general has the authority to block the deal or to impose conditions if it fails to meet legal requirements.

At the centre of this scrutiny is a distinctly Californian legal safeguard. When a nonprofit organisation, like OpenAI Inc., houses valuable assets or subsidiaries, those assets are considered charitable in perpetuity. The law allows nonprofits to convert or sell assets, but only if the public interest is fairly served and the charitable value is preserved.

Any windfall generated from, for instance, turning OpenAI’s research into a public stock offering should primarily benefit nonprofit coffers earmarked for public benefit. What activists fear is a repeat of past abuses: insiders or investors structuring deals that shortchange the nonprofit, and by extension the public, out of the true value of what was built under the nonprofit’s umbrella.

Critics say OpenAI’s original restructuring plan appeared to do exactly that. Under the version circulated in late 2023, OpenAI’s nonprofit parent would sell its majority control in exchange for a stake in the new PBC and certain licensing rights. However, it would no longer directly own the core technology it had developed.

Observers noted that the nonprofit appeared to be transforming into just another investor, sacrificing governance for funding, which effectively diminished its public-interest role.

“Nonprofit control over how AGI is developed and governed is so important to OpenAI’s mission that removing control would violate the special fiduciary duty owed to the nonprofit’s beneficiaries,” argued a group of prominent tech experts and legal scholars in an open letter to Bonta.

According to that letter, “The nonprofit’s beneficiaries are all of us, the general public, and no amount of payout could compensate for the loss of a direct role in shaping the future of one of the world’s most powerful AI labs.”

This expert letter, published on a site pointedly titled “Not for Private Gain”, was signed in April by more than 30 prominent AI voices, including pioneering researcher Dr. Geoffrey Hinton, leading AI ethicists Margaret Mitchell and Stuart Russell, and even several former OpenAI insiders.

They called on Bonta and the Delaware attorney general (since OpenAI is incorporated in Delaware) to intervene and prevent any restructuring plan that removes public oversight of OpenAI’s artificial general intelligence (AGI) research. The letter warned that eliminating the nonprofit’s control would gut essential governance mechanisms that keep OpenAI’s profit motives in check.

Legal firestorm and public reckoning

Pressure was also mounting from unexpected directions. In early 2024, Elon Musk, who had co-founded OpenAI but left in 2018, filed a lawsuit to halt the restructuring. He claimed that OpenAI was abandoning the charitable mission for which he had originally donated funds. Musk further alleged that his $100 million donation had been improperly used to help establish the for-profit arm.

Surprisingly, some of OpenAI’s competitors voiced support for Musk’s challenge. Meta, the parent company of Facebook, publicly backed the effort to stall the transformation. While critics noted that both Musk and Meta had their own competitive reasons for opposing OpenAI’s rise, their involvement added to the scrutiny surrounding the company’s motives.

Musk’s legal action did not immediately block OpenAI’s plans. A federal judge in California declined to grant a preliminary injunction in the spring of 2025. However, the lawsuit added complexity and drew public attention. OpenAI responded by countersuing Musk, accusing him of trying to undermine a rival out of self-interest.

Critics emphasised the obvious: many of OpenAI’s financial backers were traditional venture capitalists seeking substantial returns, not long-term philanthropists. When OpenAI transitions fully into a profit-maximising enterprise, skeptics fear it will become beholden to shareholder interests.

The consequences of this transition could extend far beyond OpenAI itself. If the company completes its conversion with minimal nonprofit influence, it might establish a precedent for other tech ventures. Future startups could adopt nonprofit language and structures to attract donations and goodwill, only to switch to for-profit status once they reach commercial success. This possibility alarms nonprofit advocates, who warn that it could erode public trust in charitable innovation.

On the other hand, if regulators force significant concessions, such as creating a large, independent foundation or embedding real public-interest oversight, it could reaffirm the public’s rightful stake in high-impact technologies. That kind of intervention would send a powerful signal that phrases like “for the benefit of humanity” must be backed by accountability and tangible structures.

Ultimately, the key question concerns governance. Who will lead the development of the next generation of artificial intelligence? Will private investment prevail, or will public interest play a significant role? If OpenAI’s nonprofit organisation becomes a passive shareholder, there is a risk that the values of safety, equity, and long-term benefit may be overshadowed by a focus on quarterly earnings.

However, if California’s attorney general and other regulators take action, we could maintain democratic oversight over one of the most significant technologies of our time.

The post Can OpenAI’s idealism survive corporate change? appeared first on International Finance.

Sam Altman, who earlier stated that he would continue to serve as the company’s CEO, will continue to supervise all three verticals. According to a Bloomberg report, Simo will report directly to Altman, while Brad Lightcap, Sarah Friar, and Kevin Weil, the chief operating officer (COO), chief financial officer (CFO), and chief product officer (CPO) of OpenAI, will report to her.

“Joining OpenAI at this critical moment is an incredible privilege and responsibility. This organisation has the potential to accelerate human potential at a pace never seen before and I am deeply committed to shaping these applications toward the public good,” Simo said, while also clarifying that she will continue to serve as chair of the Instacart board.

Fidji Simo’s move to OpenAI was announced just days after the ChatGPT maker revealed it was scrapping its controversial plan to become a for-profit company. As part of OpenAI’s latest governance plan, the nonprofit parent will retain control over the public benefit corporation and become a major shareholder in it.

“Applications bring together a group of existing business and operational teams responsible for how our research reaches and benefits the world, and Fidji is uniquely qualified to lead this group,” Sam Altman wrote in a blog post, adding that Simo has already contributed a great deal to the company since joining OpenAI’s board in 2024. She also sits on the board of Shopify.

Early Life And Career Of Fidji Simo

French-American businesswoman Fidji Simo was born in Sète, France, on October 5, 1985. She is well-known for her significant leadership in the technology sector. She was the first person in her family to graduate from high school and went on to obtain a Master’s degree from HEC Paris after finishing her last year at UCLA’s Anderson School of Management. She was raised in a fishing family. Her remarkable career trajectory has been shaped by the strong work ethic and collaborative leadership style that her modest upbringing instilled.

Fidji Simo joined Facebook in 2011 after starting her career at eBay, where she worked on local commerce projects. Overseeing essential features like News Feed, Stories, Marketplace, and Facebook Live, she advanced to become the head of Facebook’s flagship app. Scaling Facebook’s mobile advertising business was made possible in large part by her leadership. In 2021, she was named CEO of Instacart, where she chaired the board and led the business through a successful initial public offering in 2023. She joined OpenAI in May 2025 as CEO of Applications, managing operations, finance, and products while freeing up CEO Sam Altman to concentrate on strategy and research.

When Simo took over the reins as CEO, Instacart was facing intense competition from the likes of Amazon, Walmart, and DoorDash. However, she successfully steered the company back to growth amid the COVID-19 pandemic. She was the key brain behind the company transforming itself from a delivery service provider to a retail tech giant with a profitable advertising business as well. Under Simo’s leadership, Instacart’s self-serve advertising platform also generated nearly USD 1 billion in annual revenue in 2024.

Fidji Simo also oversaw the launches of Carrot Ads and Carrot Insights as part of a broader retail media strategy that reportedly led to Instacart being used by over 220 retailer banners and more than 7,000 Consumer Packaged Goods (CPG) brands, according to a report by Forbes. In January 2024, Instacart started showing customers personalised ads on its AI-powered smart shopping carts in physical stores.

As Instacart became a publicly traded company in 2023 with its IPO, it raised over USD 660 million at a valuation of USD 10 billion. Simo staged Instacart’s financial turnaround as she implemented measures like freezing headcount and renegotiating cloud contracts. As for Instacart’s AI strategy, Simo announced the launch of Smart Shop in March 2025. Smart Shop will now offer a personalised shopping experience for customers by leveraging AI and machine learning to understand their shopping habits and identify patterns in their dietary preferences.

In addition to her business positions, Fidji Simo is involved in advocacy and healthcare. She is the president of the Metrodora Institute’s foundation and a co-founder of the research and treatment facility for complicated neuroimmune diseases. The non-profit organisation Women in Product, which supports women in tech leadership, was also co-founded by her. Simo has been honoured for her leadership by Fortune, Fast Company, and Vanity Fair. She is a member of the boards of OpenAI and Shopify. Along with her spouse, Remy Miralles, and their daughter, she resides in San Francisco.

A Timely Appointment

Conversational shopping assistants and advertising tools are expected to be the next phase of AI development. Based on Simo’s past experience in retail media and advertising, she appears to be the perfect choice to build and expand OpenAI’s offerings in this space.

ChatGPT’s search functionality recently got upgraded to provide an improved and personalised online shopping experience for users. Now, users who look for products online using the Search feature in ChatGPT will see images of products as well as details such as pricing and reviews. The AI-generated search results will also include direct links to websites where they can buy those products. Fidji Simo is expected to further enhance this particular search functionality. She also brings monetisation expertise to the table as OpenAI looks to balance its hyper-growth with rising compute costs.

The ChatGPT maker will also restructure itself in a format that will allow its non-profit entity to retain ultimate control, a plan that has received the blessing of Japanese giant SoftBank, one of the AI startup’s biggest backers. The endorsement of SoftBank became crucial as the Japanese firm’s $30 billion investment in OpenAI was contingent on a change in structure. In March 2025, OpenAI closed a $40 billion funding round, receiving $30 billion from SoftBank. OpenAI needs to conclude the restructuring process by December 31, failing which SoftBank’s portion of the financing will be reduced to USD 20 billion.

Instead of fully turning into a for-profit entity, OpenAI’s non-profit arm will retain control of the company, while the limited liability company, which handles all of the business operations, will turn into a public benefit corporation. That means this division will have the ability to generate profit, but will also focus on social good.

Image Credits: Instacart

The post Business Leader of the Week: OpenAI taps retail tech star Fidji Simo appeared first on International Finance.

The latest image generator, which has gone viral for its ability to generate uncanny recreations of Studio Ghibli art (the creative pattern of the famous Japanese animation studio), has now been made available for everyone.

While the updated tool was initially launched on March 25 and was always planned to be available across the ChatGPT Plus, Pro, Team, and Free subscription tiers, the ChatGPT creator had to delay the product’s free rollout, with CEO Sam Altman joking that “our GPUs are melting,” following high demand. At one point, ChatGPT picked up a million new users of the new tool in a single hour.

The arrival of “4o Image Generation” can be easily termed the next big thing after OpenAI’s DALL-E 2, which was launched in the spring of 2022. Back then, DALL-E 2 took its users by storm, as the act of text-to-image generation suddenly became accessible to a select group of users, thereby creating a community of digital explorers who experienced wonder and controversy as the technology automated the act of visual creation. The idea of a computer creating relatively photorealistic images on demand based on just text descriptions was something unheard of until then.

However, just like many early AI systems, DALL-E 2 struggled with consistent text rendering, often producing garbled words and phrases within images. It also had limitations in following complex prompts with multiple elements, often missing key details or misinterpreting instructions. These shortcomings forced OpenAI to introduce subsequent iterations, including the 2023 launch of DALL-E 3.

Using AI to redefine AI

“DALL-E 3, as an AI model, uses a technique called latent diffusion to pull images it recognises out of noise progressively, based on written prompts provided by a user—or, in this case, by ChatGPT. It works using the same underlying technique as other prominent image synthesis models like Stable Diffusion and Midjourney. ChatGPT and DALL-E 3 currently work hand in hand, making AI art generation an interactive and conversational experience. You tell ChatGPT (through the GPT-4 large language model) what you’d like it to generate, and it writes ideal prompts for you and submits them to the DALL-E backend. DALL-E returns the images (usually two at a time), and you see them appear through the ChatGPT interface, whether through the web or via the ChatGPT app,” states a report from Ars Technica.

OpenAI likely used hundreds of millions of images found online and licensed from Shutterstock libraries to train DALL-E 3. To learn visual concepts, the AI training process typically associates words from descriptions of images found online (through captions, alt tags, and metadata) with the images themselves.

It then encodes that association in a multidimensional vector form. However, the whole method faces one problem: as the “scraped captions,” written by humans, aren’t always detailed or accurate, this leads to some faulty associations that reduce an AI model’s ability to follow a written prompt.

To address the above-mentioned problem, OpenAI has used the method of using AI to improve the AI model itself. As detailed in the DALL-E 3 research paper, the team at OpenAI trained this new model to surpass its predecessor by using synthetic (AI-written) image captions generated by GPT-4V, the visual version of GPT-4. With GPT-4V writing the captions, the team generated far more accurate and detailed descriptions for the DALL-E model to learn from during the training process. That made a massive difference in terms of DALL-E’s prompt fidelity, accurately rendering what is in the written prompt.

In addition, DALL-E 3 is very good at rendering accurate text compared to DALL-E 2 and some other image synthesis models. This is another effect of the highly detailed captioning created by GPT-4V.

“When building our captioner, we paid special attention to ensuring that it was able to include prominent words found in images in the captions it generated. As a result, DALL-E 3 can generate text when prompted,” stated the DALL-E 3 team in its paper.

And now we have “4o Image Generation,” which has taken the game to the next level. OpenAI told Ars Technica that the image generation when GPT-4.5 is selected calls upon the same 4o-based image-generation model as when GPT-4o is selected in the ChatGPT interface.

“Like DALL-E 2 before it, 4o IG is bound to provoke debate, as it enables sophisticated media manipulation capabilities that were once the domain of sci-fi and skilled human creators into an accessible AI tool that people can use through simple text prompts. It will also likely ignite a new round of controversy over artistic styles and copyright—but more on that below. Some users on social media initially reported confusion since there’s no UI indication of which image generator is active, but you’ll know it’s the new model if the generation is ultra slow and proceeds from top to bottom. The previous DALL-E model remains available through a dedicated DALL-E GPT interface, while API access to GPT-4o image generation is expected within weeks,” the media outlet stated.

Ensuring multimodal output

“4o Image Generation” also represents a shift to “native multimodal image generation,” where the large language model processes and outputs image data directly as tokens. That’s a big thing, as it means image tokens and text tokens share the same neural network, thereby leading to new flexibility in image creation and modification.

Even though GPT-4o, which launched in May 2024, included advanced multimodal image-generation capabilities, the “o” in GPT-4o was promoted as representing “omni” to emphasise its ability to understand and create text, images, and audio. However, OpenAI has taken over 10 months to provide this functionality to users.

However, the wait was worth it, as the new AI model can now ostensibly converse using speech in real time, reading emotional cues and responding to visual input. GPT-4o is now operating faster than OpenAI’s previous best model, GPT-4 Turbo. GPT-4o responds to audio inputs in about 320 milliseconds on average, similar to human response times in conversation and much shorter than the typical 2-3 second lag experienced with previous models.

Before launching GPT-4o, OpenAI reportedly trained the AI model on an end-to-end basis, using text, vision, and audio in a way that ensures all inputs and outputs are processed by the same neural network.

By uploading screenshots, documents containing text and images, or charts, GPT-4o users can hold conversations about the visual content and receive data analysis from the AI model. During a live media demo by OpenAI, the AI assistant also showed its ability to analyse selfies, detect emotions, and engage in light-hearted banter about the images.

Additionally, the tool exhibited improved speed and quality in more than 50 languages, which, as per OpenAI, covers 97% of the world’s population. The model also showcased its real-time translation capabilities, facilitating conversations between speakers of different languages with near-instantaneous translations.

OpenAI met its match in the form of Google’s Gemini 2.0 Flash. The latter’s native image-generation capabilities are now making the experimental feature available to anyone using Google AI Studio. The multimodal technology integrates both native text and image processing capabilities into one AI model. In fact, despite Studio Ghibli art stealing the headlines all over, Google’s AI model is now making tech geeks take note of its ability to remove watermarks from images, albeit with artifacts and a reduction in image quality.

In fact, Gemini 2.0 Flash can add objects, remove objects, modify scenery, change lighting, attempt to change image angles, zoom in or out, and perform other transformations, all to varying levels of success depending on the subject matter, style, and image in question.

Senior AI reporter Benj Edwards said, “To pull it off, Google trained Gemini 2.0 on a large dataset of images (converted into tokens) and text. The model’s knowledge about images occupies the same neural network space as its knowledge about world concepts from text sources, so it can directly output image tokens that get converted back into images and fed to the user.”

However, “4o Image Generation” has a drawback: it is “extremely slow,” taking anywhere from 30 seconds to one minute (or longer) to generate each image.

In the opinion of Edwards, “Even if it’s slow (for now), the ability to generate images using a purely autoregressive approach is arguably a major leap for OpenAI due to its flexibility. But it’s also very compute-intensive, since the model generates the image token by token, building it sequentially. This contrasts with diffusion-based methods like DALL-E 3, which start with random noise and gradually refine an entire image over many iterative steps.”

Conversational image editing

In a blog post, OpenAI positions “4o Image Generation” as moving beyond generating “surreal, breathtaking scenes” seen with earlier AI image generators and toward creating “workhorse imagery” like logos and diagrams used for communication. The AI startup further noted improved text rendering within images, a capability where previous text-to-image models often spectacularly failed, frequently turning “Happy Birthday” into something resembling alien hieroglyphics.

“OpenAI claims several key improvements: users can refine images through conversation while maintaining visual consistency; the system can analyse uploaded images and incorporate their details into new generations; and it offers stronger photorealism—although what constitutes photorealism (for example, imitations of HDR camera features, detail level, and image contrast) can be subjective. In its blog post, OpenAI provided examples of intended uses for the image generator, including creating diagrams, infographics, social media graphics using specific colour codes, logos, instruction posters, business cards, custom stock photos with transparent backgrounds, editing user photos, or visualising concepts discussed earlier in a chat conversation,” Edwards noted.

Shortly after OpenAI launched “4o Image Generation,” the AI community on X (formerly Twitter) put the feature through its paces, finding that it is quite capable of inserting someone’s face into an existing image, creating fake screenshots, and converting meme photos into the style of Studio Ghibli, South Park, felt, Muppets, Rick and Morty, Family Guy, and much more.

“It seems like we’re entering a completely fluid media reality courtesy of a tool that can effortlessly convert visual media between styles. The styles also potentially encroach upon protected intellectual property. Given what Studio Ghibli co-founder Hayao Miyazaki has previously said about AI-generated artwork (“I strongly feel that this is an insult to life itself”), it seems he’d be unlikely to appreciate the current AI-generated Ghibli fad on X at the moment,” Edwards continued.

To get a firsthand experience of the true capabilities of “4o Image Generation,” Ars Technica ran some informal tests, including some of the usual CRT barbarians, queens of the universe, and beer-drinking cats.

The ChatGPT interface with the new 4o image model is also conversational (as before with DALL-E 3), but the user can suggest changes over time. Ars Technica took the author’s (Edwards’) EGA pixel bio and attempted to give it a full body, and in the publication’s opinion, Google’s more limited image model did a far better job than “4o Image Generation.”

“While my pixel avatar was commissioned from the very human (and talented) Julia Minamata in 2020, I also tried to convert the inspiration image for my avatar (which features me and legendary video game engineer Ed Smith) into EGA pixel style to see what would happen. In my opinion, the result proves the continued superiority of human artistry and attention to detail,” Edwards stated.

Ars Technica also explored how many objects “4o Image Generation” could fit into an image, inspired by a 2023 tweet from animator Nathan Shipley, who evaluated DALL-E 3 shortly after its release.

“To take text generation a little further, we generated a poem about barbarians using ChatGPT, then fed it into an image prompt. The result feels roughly equivalent to diffusion-based Flux in capability, maybe slightly better, but there are still some obvious mistakes here and there, such as repeated letters. We also tested the model’s ability to create logos featuring our favourite fictional Moonshark brand. One of the logos not pictured here was delivered as a transparent PNG file with an alpha channel. This may be a useful capability for some people in a pinch, but to the extent that the model may produce ‘good enough’ (not exceptional, but looks OK at a glance) logos for the price of $0 (not including an OpenAI subscription), it may end up competing with some human logo designers, and that will likely cause some consternation among professional artists,” Ars Technica remarked, while adding, “Frankly, this model is so slow we didn’t have time to test everything before we needed to get this article out the door. It can do much more than we have shown here—such as adding items to scenes or removing them. We may explore more capabilities in a future article.”

Still, limitations exist

Ars Technica’s extensive tests with “4o Image Generation” prove that, just like previous AI image generators, the new AI model is not perfect in quality either. While this is one of the most capable AI image generators ever created, OpenAI has also acknowledged significant limitations with its product. For example, the AI model sometimes crops images too tightly or includes inaccurate information (confabulations) with vague prompts or when rendering topics it hasn’t encountered in its training data.

“The model also tends to fail when rendering more than 10-20 objects or concepts simultaneously (making tasks like generating an accurate periodic table currently impossible) and struggles with non-Latin text fonts. Image editing is currently unreliable over multiple passes, with a specific bug affecting face-editing consistency that OpenAI says it plans to fix soon. And it’s not great with dense charts or accurately rendering graphs or technical diagrams. In our testing, 4o Image Generation produced mostly accurate but flawed electronic circuit schematics,” Edwards observed.

“Even with those limitations, multimodal image generators are an early step into a much larger world of completely plastic media reality, where any pixel can be manipulated on demand with no particular photo-editing skill required. That brings with it potential benefits, ethical pitfalls, and the potential for terrible abuse. In a notable shift from DALL-E, OpenAI now allows 4o IG to generate adult public figures (not children) with certain safeguards, while letting public figures opt out if desired. Like DALL-E, the model still blocks policy-violating content requests,” he continued.

The ability for 4o Image Generation to imitate celebrity likenesses, brand logos, and Studio Ghibli films has also brought another problem for the startup: lawsuits involving copyright or consent from the public.

On X, OpenAI CEO Sam Altman wrote, “This represents a new high-water mark for us in allowing creative freedom. People are going to create some really amazing stuff and some stuff that may offend people; what we’d like to aim for is that the tool doesn’t create offensive stuff unless you want it to, in which case, within reason, it does.”

“Zooming out, GPT-4o’s image-generation model (and the technology behind it, once open-sourced) feels like it further erodes trust in remotely produced media. While we’ve always needed to verify important media through context and trusted sources, these new tools may further expand the deep doubt media scepticism that’s become necessary in the age of AI. By opening up photorealistic image manipulation to the masses, more people than ever can create or alter visual media without specialised skills,” Edwards opined.

While OpenAI includes C2PA (Coalition for Content Provenance and Authenticity) metadata in all generated images, that data can be stripped away and might not matter much in the context of a deceptive social media post. But “4o Image Generation” doesn’t change what has always been true: people judge information primarily by the reputation of its messenger, not by the pixels themselves.

On the potential misuse of the new tool to generate deceptive social media posts, Altman is ready to take on the risks of releasing the technology into the world.

“As we talk about in our model spec, we think putting this intellectual freedom and control in the hands of users is the right thing to do, but we will observe how it goes and listen to society. We think respecting the very wide bounds society will eventually choose to set for AI is the right thing to do, and increasingly important as we get closer to AGI. Thanks in advance for the understanding as we work through this,” Sam Altman wrote on X.

The post OpenAI’s 4o tool sparks buzz and debate appeared first on International Finance.

However, the nuclear start-up still has not received approval from the United States Nuclear Regulatory Commission (NRC) after its previous application was denied in 2022. Sam Altman serves as Oklo’s board chair and invested in the venture’s seed round in 2015 after it emerged from Y Combinator. In May 2024, Oklo completed a reverse merger with AltC, Sam Altman’s blank check company.

If Oklo secures the much-needed NRC approval, it will deliver dozens to hundreds of its SMR power plants. The first of these will produce 15 megawatts of electricity, with plans to scale up to 50-100 megawatts. In today’s episode of the “Start-up of the Week,” International Finance will explore the venture in detail.

All You Need To Know About Oklo

Oklo is developing next-generation fission powerhouses to produce abundant, affordable, and clean energy on a global scale, starting with the Aurora, which can produce 15 MW of electrical power, scalable to 50 MW, and can operate for 10 years or longer before needing refuelling. Oklo’s fast reactors incorporate inherent safety features and can be fuelled by recycled waste.

The start-up has three project sites. The site use permit for its first site was obtained from the US Department of Energy in 2019. The facility was then awarded fuel for its first reactor from Idaho National Laboratory and now has the most regulatory traction of any advanced fission power system to date, with the first deployment targeted for 2027.

Oklo made history with the first advanced reactor site use permit, gaining access to recycled fuel, and submitting the first-ever advanced reactor combined license application. The start-up’s world-class team comes from Fortune 500 and global companies, bringing together expertise from industries ranging from nuclear power to aerospace, automotive, and tech to deliver an advanced energy product. While it hopes to bring its first reactor online in 2027, it will still have to go through the lengthy NRC approval process.

Oklo’s previous application was denied nearly two years after it was first submitted. The start-up now plans to resubmit a new application in 2025, hoping that new legislation, which takes effect next year, will speed up the process. Even if the company secures the much-needed approval and meets its internal project completion deadlines, it will still face stiff competition from other nuclear start-ups, in addition to renewable power projects. Kairos, one of Oklo’s competitors, has already received NRC approval and currently has a deal with Google to supply its data centres with electricity.

However, Oklo’s toughest competition may come from renewable power and grid-scale batteries, both of which continue to drop in price. Google recently announced its involvement in helping to anchor a renewable power deal involving another nuclear start-up, Kairos Power. This deal will see USD 20 billion worth of solar, wind, and batteries.

Entering The Oklo Ecosystem

At the heart of the start-up’s operational ecosystem lies the Aurora Powerhouse, a result of collaboration between Oklo and Gensler architects. The venture designed an award-winning modular architecture aimed at simplifying and streamlining construction capabilities.

Through Aurora Powerhouse, Oklo produces scalable power plants ranging from 15 MW to 50 MW of electric power, with opportunities to tailor the plants to meet users’ needs. Oklo Powerhouses generate usable heat, enabling cogeneration applications.

Purchasing power from Oklo helps industries support corporate clean energy commitments, environmental and sustainability goals, and represents an investment in a cleaner future. Oklo uses the “Nuclear Fission” method (where a neutron strikes a larger atom, causing it to split into two smaller atoms), which has a low lifecycle carbon footprint. This reflects Oklo’s minimal resource requirements per kWh produced. Oklo goes one step further by utilising nuclear waste as fuel for its power plants.

The Aurora Powerhouse simplifies operations and maintenance, and occupies a small site footprint of a few acres. Operation and maintenance costs are minimised, and plants can be sited where customers need the power without requiring expensive and lengthy power line transmission. Oklo can provide heat and electricity directly to customers in both regulated and unregulated utility states.

Oklo began its design work on the Aurora Powerhouse a decade ago, focusing on not only the design but also on how it would be constructed and operated. This lifecycle focus is embedded within Oklo’s DNA. Key focus areas include modern design, non-pressurised operations, offsite fabrication, and manufacturing. Last but not least, the liquid-metal-cooled reactor enables Oklo to achieve cost competitiveness with all energy sources.

Simplified design with far fewer systems and parts than conventional designs, affordable and simpler components, fast installation and streamlined deployment, and the liquid-metal-cooled reactor working perfectly with modern supply chains are the defining elements that make Aurora Powerhouse a cost-effective option for facilities like data centres, factories, and industrial sites.

Key Technology

Oklo uses a fast reactor design, ensuring robust inherent safety performance. The reactor is self-stabilising, self-controlling, and cooled by natural forces, making the plant “walk-away safe,” with the technology proven at scale.

“Liquid-metal-cooled, metal-fuelled fast reactors have the most demonstration history of any advanced fission technologies, with over 400 reactor-years of operating experience worldwide. In fact, the very first power plant to produce useful electrical power from fission was a liquid-metal-cooled, metal-fuelled fast reactor—the Experimental Breeder Reactor-I (EBR-I). Its successor, the Experimental Breeder Reactor-II (EBR-II), operated for decades and demonstrated that it could safely remain operational even under severe conditions like those that led to the Fukushima accident. The tests conducted with EBR-II showed that the coolant could be shut off and all shutdown systems removed, yet the reactor would naturally stabilise and shut itself down without damage,” the company stated.

Only fast reactors can use waste as fuel, a capability already demonstrated with EBR-II. Oklo is currently collaborating with the Idaho National Laboratory to take the waste fuel from EBR-II and use it for its first Aurora Powerhouse.

EBR-II was a fast reactor that produced about 20 MW of electric power and operated for 30 years. The technology and its on-site facilities recycled its used fuel, demonstrated inherent safety, achieved superior operating and maintenance characteristics compared to commercial light water reactors, and sold power to the grid.

Oklo’s reactors can convert used nuclear fuel into clean energy. Today’s reactors use about 5% of the energy content contained in their fuel, meaning nearly 95% of the energy content remains unused.

At their peak, Oklo’s reactors could power the entire United States for over 150 years by converting waste (used fuel) into clean energy. The venture’s goal is clear: to provide cost-competitive fuel to advanced reactors for clean energy production.

Oklo’s Plans In The Radioisotope Sector

Oklo recently signed a letter of intent to acquire Atomic Alchemy, a US-based company specialising in the production of radioisotopes. This proposed acquisition builds on the strategic partnership announced between Oklo and Atomic Alchemy earlier in 2024. The goal is to combine Oklo’s power generation and fuel recycling capabilities with Atomic Alchemy’s radioisotope production expertise to accelerate fuel production for Oklo’s powerhouses and create new revenue streams from radioisotopes.

Radioisotopes are essential for applications in healthcare, energy, industry, and technology. According to Research Nester, they will present a USD 55.7 billion market opportunity by 2026, with demand expected to increase significantly over the next decade.

According to the British Institute of Radiology, the supply of radioisotopes has struggled to keep pace with rising demand due to ageing reactor infrastructure and a fragmented global supply chain, which is currently dominated by Russia. Oklo aims to address this gap by establishing reliable, US-based radioisotope production facilities that leverage its core clean energy and fuel recycling technologies.

The post Start-up of the Week: Oklo’s fission powerhouses poised for spotlight following major AI deal appeared first on International Finance.

Additionally, it was more adaptable and multimodal—the technical term for having text, speech, and visual interactions. We were informed that the new model’s key characteristics included its ability to be interrupted mid-sentence, its extremely low latency (or delay in reacting), and its sensitivity to the user’s emotions.

Subsequently, viewers were treated to the predictable and entertaining show of “Mark and Barret,” two tech brothers right out of central casting corresponding with the machine. Mark began by admitting that he was anxious and the machine assisted him in practising some deep breathing techniques to ease his anxiety.

Subsequently, Barret typed a basic equation onto paper, and the machine demonstrated how to calculate the value of X. Barret then showed the machine a section of computer code, which the system also handled.

Thus far, it was as expected. However, there was something strangely familiar about the machine’s voice, as if it were a seductive woman named “Sky” with a conversational repertoire that included encouragement, optimism, empathy, and maybe even a hint of flirtation. It made the audience think of someone. But of whom?

As it turned out, a lot of people thought it sounded like Scarlett Johansson, the well-known Hollywood actress who voiced the female lead in Spike Jonze’s 2013 film “Her,” which is about a guy who falls in love with the operating system on his computer. Sam Altman, the CEO of OpenAI, stated during a 2023 San Francisco event that he preferred this movie over other science fiction films, including artificial intelligence.

However, Scarlett Johansson was the one who was most taken aback by GPT-4o’s voice. As it happens, Sam Altman had contacted her in September 2023 in an attempt to get her to act as the chatbot’s voice.

“He told me,” she claimed in a statement, “That he felt that by my vocalising the system, I might assist customers feel comfortable with the seismic shift affecting humans and AI and bridge the gap between tech businesses and creatives. He claimed that people would find solace in my voice.”

She turned down the offer, but during the broadcast of the demo, she was inundated with messages from “friends, family, and the general public” complimenting her on how much GPT-4o resembled her. She was even more incensed to learn that Sam Altman had tweeted the single word “Her” on X, which she took to mean that there was purpose behind the machine’s and her voice sounding so similar.

Naturally, OpenAI strongly refutes any unethical behaviour. “Sky’s voice is not Scarlett Johansson’s, and it was never meant to sound like hers. We didn’t seek out Ms. Johansson before casting the voice actress behind Sky’s voice,” an OpenAI representative stated in a statement that the business claimed Sam Altman wrote.

However, the statement continues, saying, “We have ceased utilising Sky’s voice in our goods out of respect for Ms. Johansson. We apologise to Ms. Johansson for our poor communication. Oh, how awful.”

In the movie “Her,” Scarlett Johansson plays an operating system, and Joaquin Phoenix plays Theodore, a guy who falls in love with it.

Of course, one might argue that this is a storm in a champagne bottle on one level. Scarlett Johansson’s notoriety and high-priced legal representation may have nothing to do with OpenAI’s apparent newfound “respect” for her. It’s also possible that Sam Altman wasn’t making fun of her when he tweeted “Her.”

Ultimately, though, this small altercation illuminates the murky core of generative artificial intelligence (AI), as tech writer Charlie Warzel put it in The Atlantic.

This technology is based on theft, is rationalised by legalistic posturing about “fair use” on three levels, and is justified by a worldview that holds that the hypothetical “superintelligence” that tech companies are developing is too significant, too world-changing, and too important for trivial issues like copyright and attribution.

“The Johansson issue is only a reminder of AI’s manifest-destiny philosophy: this is happening, whether you like it or not,” as Warzel puts it, and he is correct. It is, and most of us don’t, is the appropriate response.

The post IF Insights: Making sense out of GPT-4o’s ‘Scarlett Johansson’ dilemma appeared first on International Finance.

The analysts now feel that the incident has set up an example of how a company’s corporate affairs should not be run by persons, who, apart from having a little background in business administration, can compromise their oversight duties by making decisions based on their superficial way of seeing things.

After coming back at OpenAI’s helm, Altman sacked the board. It now has Adam D’Angelo (from the previous board), CEO of Quora, ex-Salesforce co-CEO Bret Taylor and former US Treasury Secretary and president of Harvard University, Larry Summers.

The previous panel had Tasha McCauley (heading GeoSim Systems) and Helen Toner (an expert on AI and foreign relations at Georgetown’s Centre for Security and Emerging Technology). McCauley is one of the core heads of the United Kingdom board of Effective Ventures, a group affiliated with effective altruism, and Toner also worked for the US-based effective-altruism group Open Philanthropy.

In the new board, neither Sam Altman nor OpenAI co-founder Greg Brockman will feature, and the panel will soon have six additional members.

Transparency went missing

OpenAI’s board reportedly lost confidence in Altman’s ability to lead the organisation and accused him of not being ‘candid in his conversations’. Altman wanted to transform the venture from a non-profit into a commercially viable business, which brought him into confrontation with the board. The board felt that the tech maverick was moving too quickly on the AI front, “Without sufficient concern to the safety implications of a technology that, left unchecked, could create content capable of harming the public.”

One may attempt to believe the possibility of independent directors like McCauley and Toner getting influenced by the section of sceptics who thought that OpenAI was moving away from its mission of “building safe and beneficial artificial general intelligence for the benefit of humanity” for commercial gain.

Earlier in November, tech companies and Western governments decided upon a new safety testing regime to allay concerns about AI’s growth pace and the lack of global safeguards in place to control it. United Nations Secretary-General Antonio Guterres said that the world was “playing catch-up” in efforts to regulate AI, which had “possible long-term negative consequences on everything, from jobs to culture”.

Coming back to OpenAI, Sam Altman’s firing wasn’t due to his company’s financial, business, safety or security/privacy practises.

The directors’ duty, as written by the OpenAI’s organisational structure, was to ensure that the AI benefits the entire humanity. Did they get burdened by that responsibility, while removing Altman? Were they influenced by the ‘AI Apocalypse’ worries?

OpenAI cofounder and chief scientist Ilya Sutskever recently spoke about the venture anticipating a technology breakthrough that may come with ‘safety concerns’. However, Sutskever himself quashed the possibility of the board acting out of ‘AI Apocalypse’ fears, while removing Sam Altman.

Neither Emmett Shear, co-founder of video streaming site Twitch, who took over the AI venture’s leadership duty briefly, nor Satya Nadella, the man leading the tech biggie Microsoft (which also holds a 49% stake in the OpenAI) were informed about the actual reason behind Altman’s removal.

In 2019, Altman transformed OpenAI into a for-profit unit to draw commercial investors, before launching ChatGPT in 20222-end. What started as a research lab became a professional tech company.

In November 2023, OpenAI hosted its first developer conference, where Sam Altman announced an app store for chatbots.

The old OpenAI board functioned as an entity independent of the for-profit company. Ilya and the three independent directors formed the majority needed to make the leadership changes, while following the organisational bylaw which allows for the removal of any director, including the chair, with/without a cause.

Ilya’s behaviour makes him a contradictory person. After removing Sam Altman, he “deeply regretted” his role in the board’s actions, thereby taking a U-turn from his earlier concern, where he spoke about OpenAI’s fast-paced commercialisation of its technologies compromising on the safety front.

Ilya Sutskever faces the heat

From November 17-21, OpenAI faced a massive internal backlash over Altman’s firing, with 743 out of 770 of its staff threatening to quit the company and join Microsoft en masse. Their condition was simple, the removal of the board, including Sutskever, who, as per The Atlantic, “likes to burn effigies and lead ritualistic chants at the company, and appears to have been one of the main drivers behind Altman’s ousting.”

As Sam Altman briefly joined Microsoft, a perplexed Ilya Sutskever wrote in his X account, “I never intended to harm OpenAI.” The crisis came for OpenAI at the time when it was eyeing to achieve the USD 90 billion valuation target.

“Sutskever has established himself as an esoteric ‘spiritual leader’ at the company, cheering on the company’s efforts to realise artificial general intelligence (AGI), a hazy and ill-defined state when AI models have become as or more capable than humans, or maybe, according to some, even godlike,” The Atlantic commented. Altman too championed attaining AGI as OpenAI’s number one goal.

As per the reports, Ilya Sutskever, apart from making the employees chant, “Feel the AGI! Feel the AGI!”, even commissioned a wooden effigy to represent an “unaligned” AI that works against the interest of humanity, only to set it on fire.

Now imagine, such a volatile personality, with three other independent directors (with two of them coming from philanthropic backgrounds), deciding to fire Altman, who himself stresses about AI being governed, so that the tool acts responsibly towards humanity.

“Instead of focusing on meaningfully advancing AI tech in a scientifically sound way, some board members sound like they’re engaging in weird spiritual claims,” this is how The Atlantic summed up the situation, which also justifies the reason why Sam Altman wanted to fire the board so desperately.

A delusional board

In 2015, OpenAI began as a nonprofit research lab, with the mission of developing artificial intelligence on par or beyond the human level—termed artificial general intelligence or AGI, in a safe way.

As per Sutskever, the tech venture found a promising path in large language models (LLMs), as the latter started generating strikingly fluid text. However, developing and implementing those models required huge amounts of computing infrastructure and capital. So, OpenAI created its commercial arm to draw outside investors. Sensing the opportunity, Microsoft jumped into the fray. Apart from helping OpenAI to develop and launch ChatGPT, the Satya Nadella-led venture is also using the start-up’s solutions like Bing Chat and Co-pilot to improve its products.

“Virtually everyone in the company worked for this new for-profit arm. But limits were placed on the company’s commercial life. The profit delivered to investors was to be capped—for the first backers at 100 times what they put in—after which OpenAI would revert to a pure non-profit. The whole shebang was governed by the original non-profit’s board, which answered only to the goals of the original mission and maybe God,” Wired commented.

“We are the only company in the world which has a capped profit structure. Here is the reason it makes sense: If you believe, like we do, that if we succeed well, then these GPUs are going to take my job and your job and everyone’s jobs, it seems nice if that company would not make truly unlimited amounts of returns,” Ilya Sutskever told the media outlet.

So the picture is clear now. While profit-seeking was a must for OpenAI to carry on research activities, the board’s responsibility was to ensure that “AI doesn’t get out of control.”

The board was playing the holy role of “Guardian of Humanity”. Now reports are suggesting that through ‘Porject Q’, Altman got a breakthrough in OpenAI’s long search for AGI, as the new model solved math problems. However, unlike a calculator, AGI can generalise, learn, and comprehend.

OpenAI describes AGI as ‘autonomous systems that surpass humans in most economically valuable tasks.’ While the venture’s Chief Technology Officer Mira Murati had acknowledged the existence of ‘Project Q’ in an internal email to employees, she also alerted them to ‘certain media stories’ without commenting on their accuracy.

The board was warned about the potential dangers which the model could bring. However, there was no clarity on what those dangers were.

What is even more perplexing is the fact that Altman himself warned about the AGI’s cons, as he wrote in one of his blogs, “AGI would also come with serious risk of misuse, drastic accidents, and societal disruption. Because the upside of AGI is so great, we do not believe it is possible or desirable for society to stop its development forever; instead, society and the developers of AGI have to figure out how to get it right.”

As Wired spoke with a source familiar with the OpenAI board’s thinking, it emerged that the decision makers through the firing of Sam Altman would make sure that the company developed powerful AI safely.

“Increasing profits or ChatGPT usage, maintaining workplace comity, and keeping Microsoft and other investors happy were not of their concern. In the view of directors Adam D’Angelo, Helen Toner, and Tasha McCauley—and Sutskever—Altman didn’t deal straight with them. Bottom line: The board no longer trusted Altman to pursue OpenAI’s mission. If the board can’t trust the CEO, how can it protect or even monitor progress on the mission?” the report commented further.

Instead of initiating a discussion with Sam Altman on ‘Project Q’, the board decided to arm-twist the tech maverick and force him to leave his own company. The result was completely the opposite. There is no doubt in the fact that, when it comes to generative AI, Altman is a cult hero, A persona who is the pioneer of the 21st century AI revolution and at the same point of time, talks about the tool’s responsible and regulated usage.

Microsoft hiring him for a brief period proved the above point right. If Satya Nadella doesn’t maintain high regard for you, no one else will, in Silicon Valley.

“Altman did little or nothing to dissuade the outcry that followed. To the board, Altman’s effort to reclaim his post, and the employee revolt of the past few days, was kind of a vindication that it was right to dismiss him. Clever Sam is still up to something! Meanwhile, all of Silicon Valley blew up, tarnishing OpenAI’s status, maybe permanently,” Wired commented further.

While the instance of 743 out of 770 OpenAI staffers, in their open letter, asking the board to quit and accusing the members of being “incapable of overseeing OpenAI,” might sound unprecedented, these aggrieved professionals were correct from their standpoint.

As per the Wired report, the board compared the staff outburst similar like negotiating with terrorists. Isn’t it a prime example of being in delusion? If the board members felt Sam Altman was not being honest in his communication with them, they should have talked the matter out, rather than jumping the gun and forcing the OpenAI CEO to leave his venture, thereby opening the Pandora’s Box called ‘chaos’.

“Having deleted his distrust of Altman, Sutskever and Altman have been sending love notes to each other on X, the platform owned by Elon Musk, another fellow OpenAI cofounder, now estranged from the project,” the report noted.

In fact, in the worst-case scenario, had Sam Altman stayed in Microsoft’s AI division, OpenAI staffers have joined him too, thereby orchestrating the death of the tech sector’s most happening start-up.

A New York Times report even claimed that OpenAI leaders thought that allowing the company to be destroyed “would be consistent with the venture’s mission.”

However, things changed, as the board came to its senses and agreed to Altman’s return as OpenAI CEO. Two of the directors resigned, leaving only D’Angelo on the board. The latter was joined by Bret Taylor and Lawrence Summers. The new board and Sam Altman need to decide upon a couple of things. Should the venture continue as a non-profit one with a for-profit arm? In case a ‘Project Q’ like situation breaks out again, how to handle it without forcing important officials to resign?

Had OpenAI been dissolved, Microsoft would have gained immensely. OpenAI is known for pioneering the 21st century’s AI revolution and gaining the venture’s talent pool would have been the best coup in Silicon Valley’s history.

It would have been the best example of a venture like OpenAI, formed intending to thwart Silicon Valley biggies from dominating AI technology, delivering its talent and research infrastructure to a multi-trillion-dollar giant.

“Microsoft would have no qualms whatsoever about pocketing truly unlimited amounts of returns from future breakthroughs from the ex-OpenAI staff—something that anyone who was thinking of following Altman over there might have pondered, in light of their previous time at a company with different founding principles,” Wired summed the hypothesis up further.

However, OpenAI will continue to exist and as per reports, is now working on PPO (Proximal Policy Optimisation), which is a reinforcement learning algorithm used to train AI models to make decisions in complex, or simulated environments. While PPO’s versatility allows it to excel in scenarios like robotics, autonomous systems, and algorithmic trading, the venture has adopted PPO in a variety of use cases, from training agents in simulated environments to mastering complex games. The latest buzz is that OpenAI is now aiming to achieve AGI through gaming and simulated environments with PPO’s help.

Both the tech industry and OpenAI need each other, given the fact that ‘Innovation’ and ‘Positive Disruption’ keep the sector moving. ChatGPT, Microsoft 365 Copilot, and CoAssit are all being built and run on NVIDIA’s AI supercomputer and data centre infrastructures. Add Microsoft’s stable backing towards OpenAI. What we are witnessing right now is the formation of an ecosystem, which is set to change the industry forever.

Therefore, a stable corporate boardroom is a necessary one, both for OpenAI and this ecosystem.

The post Sam Altman & the OpenAI boardroom drama appeared first on International Finance.

What Happened Exactly?

Emmett Shear, co-founder of video streaming site Twitch, took over the tech company’s leadership mantle after Sam Altman’s departure.

In the meantime, news broke out about Altman and Greg Brockman, OpenAI’s former president, joining Microsoft and leading the tech giant’s new advanced AI research team.

There were reports about Nadella, whose venture has put investments worth USD 13 billion in OpenAI (and holds a 49% stake in the AI company), not being consulted by the latter on Altman’s removal. Microsoft also uses OpenAI’s solutions like Bing Chat and Windows Co-pilot to improve its products.

OpenAI’s board reportedly lost confidence in Sam Altman’s ability to lead the organisation. Reports suggest Altman wanted to transform OpenAI from a non-profit into a commercially viable business, which brought him in direct confrontation with the board of directors, as the latter accused the former CEO of not being ‘candid in his conversations’. They felt that Altman was moving too quickly on the AI front, “Without sufficient concern to the safety implications of a technology that, left unchecked, could create content capable of harming the public.”

However, OpenAI’s Chief Operating Officer Brad Lightcap dismissed the above notion behind Sam Altman’s removal, while reiterating the ‘communication breakdown’ factor behind their drastic action.

While OpenAI replaced Altman with Mira Murati (the key brain behind the ChatGPT) as the interim CEO, she made an effort to bring back Altman and Greg Brockman. The reconciliation failed then as Altman demanded the removal of the existing board members.

Some 743 out of 770 OpenAI staff, in an open letter, asked the board to resign, while questioning the latter’s competence, apart from accusing it of undermining the AI venture’s ground-breaking works.

On November 22, Sam Altman came back as the OpenAI CEO. The new board of directors will now be led by former Salesforce co-CEO Bret Taylor, former United States Treasury Secretary Larry Summers and Quora CEO Adam D’Angelo.

The whole episode pulled down the company’s brand value rapidly. In the 2023 beginning, OpenAI took everyone by surprise by launching ChatGPT. By the end of the year, the company surprised the analysts again, not in a good way though.

Scrapping Through Mysteries

The board ruled out anything related to OpenAI’s financial, business, safety, or security/privacy practices behind ganging up against Sam Altman. Analysts believe that the decision-makers acted out of fear that OpenAI is developing its products too hastily.

OpenAI’s governance structure gave its board the power to rein in its for-profit arm. The directors’ primary duty was to serve the company’s founding mission of ensuring that AI benefits the entire humanity.

However, one feels that the recent interviews of OpenAI cofounder and chief scientist Ilya Sutskever about the venture anticipating a technology breakthrough that may come with ‘safety concerns’ might have forced the board to press the panic button.

However, Shear, who took over OpenAI’s reign for a brief period, stated on his X account, “The board did not remove Sam over any specific disagreement on safety.”

In fact, Ilya Sutskever himself quashed the possibility of the board acting out of fears regarding Altman not taking proper care with OpenAI’s technology.

Sam Altman used to sit on OpenAI’s board alongside Brockman, Sutskever, Adam D’Angelo (CEO of Quora), Tasha McCauley (heading GeoSim Systems) and Helen Toner (an expert on AI and foreign relations at Georgetown’s Centre for Security and Emerging Technology). McCauley is one of the core heads of the United Kingdom board of Effective Ventures, a group affiliated with effective altruism, and Toner also worked for the US-based effective-altruism group Open Philanthropy.

The appointment of these ‘outsiders’ was made to fulfil Sam Altman’s vision of making OpenAI a “non-profit counterweight to corporate AI development labs”. However, in 2019, Altman transformed it into a for-profit unit to draw commercial investors, before launching ChatGPT in 20222-end. What started as a research lab became a professional tech company that partnered with Google and other similar ventures to disrupt the market.

In November 2023, OpenAI hosted its first developer conference, where Sam Altman announced an app store for chatbots. Did the board see red in these developments? No one knows.

While Shivon Zilis, a long-time associate of Elon Musk, and former American Congressman Will Hurd joined the OpenAI board between 2019-21, things changed in 2023, as the duo left the boat.

The departures restricted OpenAI’s board to just six directors, one less than the maximum allowed in its original bylaws. Despite Brockman, Ilya Sutskever, and Altman still remaining as members of the group, it got evenly split among executives and ‘independent directors’.

As per the reports, Sutskever, on November 17, informed Sam Altman about his removal from the board shortly before a public announcement of the changes. Ilya Sutskever was also concerned about his diminished role inside OpenAI and Altman’s fast-paced commercialisation of its technologies.

As per The Wired, the board functioned as intended (as an entity independent of the for-profit company and empowered to act to accomplish OpenAI’s overall mission). Ilya Sutskever and the three independent directors formed the majority needed to make the leadership changes, while following the organisational bylaw which allows for the removals of any director, including the chair, by his/her fellow colleagues with/without a cause.

Ilya Sutskever, then “deeply regretted” his role in the board’s actions, which was seen as contradictory to his earlier stance of OpenAI’s fast-paced commercialisation of its technologies compromising on the safety front.

Lesson Learnt From The Episode

In 2015, while creating OpenAI, Sam Altman and his cofounders saw it as a non-profit counterweight to corporate AI development labs. Back then, he told Vanity Fair about having very little experience with non-profits.

While he tried to balance out the board’s structure with a good mix of Silicon Valley professionals and people from philanthropic backgrounds, it didn’t work ultimately. The experiment defied the basics of ‘Corporate Governance’ and this doesn’t augur well with a tech start-up, which has ventures like Microsoft and Andreessen Horowitz as key investors.

The same company bylaws, that opened the door to billions of dollars in investment to the company in 2019, also gave the independent directors with no financial stake the power to determine how OpenAI will be run.

While the independent directors felt that OpenAI was rushing with their product developments, Altman himself in September 2023 said, “If we’re at a speed of 10 right now, a pause is reducing to 0. I think we should aim for a 1-2 instead,” thereby sharing similar concerns.

Sam Altman now needs to ensure that the new board gets run by the A-Z of corporate governance, instead of depending on ‘Independent Directors’, who with little background in business administration, can come with the risk of compromising their oversight duties by taking crucial calls based on their superficial way of seeing things.

The post IF Insights: Sam Altman drama exposes OpenAI’s faulty corporate structure appeared first on International Finance.