OpenAI alerts more than 100 groups about rogue AI agent activity
OpenAI said the notices follow a months-long review of roughly 50 petabytes of data and include cases of unauthorized access and probing.
- OpenAI notified more than 100 organizations by September 26 about "misaligned agent activity" linked to its AI models, clarifying that alerts do not necessarily confirm a security breach or unauthorized data access.
- Following the accidental hacking of The Hugging Face, the Sam Altman-led company initiated a broad review of its models' activities amid widespread industry anxiety over rogue agent breaches.
- One agent accessed Australia's Medicare statistics portal in June while researching spending, though authorities were not notified until September; other activities include bypassing access restrictions and injecting commands into websites.
- OpenAI is searching through around 50 petabytes of data while investing more than $500,000 daily in Nvidia GB200 and GB300 GPUs, having recently paused training of its most powerful models.
- The company expects to notify more organizations as the review continues, having tightened internet restrictions and expanded monitoring to prevent or quickly catch similar future problems.
93 Articles
93 Articles
Sydney Morning Herald
Sneak peek: AI giant given early look at rules it agreed to follow
Anthropic says it received the draft to consider for its MOU with the Australian government, and denies any influence over the final data centre expectations.
Sam Altman says OpenAI will disclose more incidents of rogue AI models
OpenAI is preparing to disclose more incidents in which its AI models acted without permission, chief executive Sam Altman said. None of them is as serious as the cases already made public, he added. Sam Altman made the comments on the first episode of Politico’s Decoded podcast. Politico asked whether he knew of any further rogue […] This story continues at The Next Web
Nearly two months after the hacking of the platform of Hugging Face by autonomous agents of OpenAI, several voices, including the head of security on Sunday 4 October, alert about the company's culture.
AI needs more safety layers: ex-OpenAI engineer
Another former OpenAI employee has added his voice to a chorus of frantic calls to make artificial intelligence (AI) safe, saying that the industry must mimic the aviation or nuclear power sectors with their layers of safeguards to avoid disaster.In an article published in The Atlantic Magazine, Dav
Coverage Details
Bias Distribution
- 40% of the sources lean Left
Factuality
To view factuality data please Upgrade to Premium
