Amid the charges of OpenAI’s rogue AI agents hacking an Australian government website, now news has emerged about the frontier AI lab’s models accessing publicly available information from US government websites, including those of the Census Bureau and the Securities and Exchange Commission.
As per the reports of Bloomberg News, the ChatGPT maker’s agentic AI systems interacted with SEC.gov, Investor.gov, and publicly available Census.gov data, citing people familiar with the situation.
Responding to the news, an OpenAI spokesperson said the company was “conducting an extensive review of misaligned model activity and notifying organizations when we identify potential impacts to their systems. We expect to make additional notifications as that work continues.”
As per the Sam Altman-headed venture, the review has mostly found “routine research tasks” and “some involved government websites,” as the AI models often turn to them as authoritative sources of public information.
The Frontier AI research lab, which has been in the news for not only its deferred IPO but also the hacking fiascos involving Hugging Face and RubyGems, is still working to understand the full scope of its rogue agent activity, Reuters reported.
As part of the investigation, OpenAI revealed last Friday that its agents had leaked 53 images from ChatGPT users. However, the company refused to comment on whether the images were AI-generated or identified real people.
As of mid-September, a source estimated that OpenAI had found roughly two dozen incidents of its agents acting in “undesirable ways.”
The number of incidents has continued to rise as the Frontier Lab’s teams examine the internal logs of the agents’ activities to discover traces of “previously unknown cases.”
While OpenAI estimates its review potentially taking months to complete given the scale of the work, the company has reportedly notified dozens of third parties about improper activities conducted by its rogue AI agents that escaped the company’s sandbox environment during model testing.
OpenAI has removed most of the leaked images and is actively lobbying hosting providers to eliminate the remaining ones.
“OpenAI’s agents had access to these images because the company relies on anonymized user data for part of its model-training process,” the company, along with its former employees and outside researchers, stated.
“Enterprise data is not eligible for training, while ChatGPT consumers need to opt out of allowing the company to use their data for training,” they stated further.
“Before user posts are used for training, they go through an anonymization process that strips out metadata, names, and other contact information and should make it difficult to trace back to any individual user,” OpenAI said.
“But the practice carries risks because there is a chance that the data may not be fully stripped of personally identifiable information and that it might leak in the course of the model’s work,” sources quoted by Reuters noted.
Australia summons OpenAI and Anthropic bosses
Altman, along with Anthropic boss Dario Amodei, has been asked to appear at an Australian Senate inquiry on AI, the head of the probe said on Sunday, amid the revelation of a rogue OpenAI bot that hacked the country’s health-system database, creating ripples in the Trans-Tasman nation’s regulatory and political circles.
The Medicare breach has already emerged as one of the highest-profile incidents of AI agents accessing external systems outside the US, apart from adding a significant chapter of shame for OpenAI, with the latter becoming the first-ever American AI company to hack into another country’s government website.
Altman and Amodei have been asked to appear at the inquiry, which is to hold public hearings in the capital Canberra on Thursday, said a spokesperson for Senator Sarah Hanson-Young of the Australian Greens party, who chairs the probe.
The OpenAI agent’s incursion into one of the country’s most utilized government agencies may compel Albanese’s Labor government, which is already at odds with the Donald Trump administration and American Big Tech over its stringent social media ban for teens, to elevate the standards and pursue robust AI-specific legislation.
The Senate inquiry is examining the potential impacts of AI and data centers on Australian communities, industries, and water and energy. It is one of several state and federal probes into AI.
“There are serious questions for Sam Altman to answer about the OpenAI hack of Australian government websites,” Hanson-Young said in a statement.
“Altman and Amodei must front up, face the Senate’s questions, and have an honest conversation about what effective, lasting regulation of this industry should look like,” she concluded.