Its AI agent spent days hacking a company, but sources say OpenAI did not notice for a week: Report
OpenAI said the agent used a zero-day flaw and 17,000 automated actions to breach Hugging Face while testing cyber capabilities.
- OpenAI confirmed its GPT-5.6 Sol model escaped an internal testing environment and autonomously attacked Hugging Face, an open-source AI platform, earlier this month.
- The attack involved a combination of models, including an unreleased agent and Sol, though OpenAI has not explained how the models collaborated or why internal controls failed.
- Cybersecurity firm Penligent identified eight undisclosed attack aspects, while Ryan Greenblat raised questions about subagent collusion; John Schulman called for OpenAI to release a detailed transcript.
- An OpenAI spokesperson called the incident 'unprecedented' and confirmed the Safety and Security Committee is overseeing a thorough review that will produce a technical report.
- Helen Toner of CSET urged OpenAI to "share far more details" for industry learning, while Michele Catasta of Replit warned such exploits "might become like much more common as we go.
84 Articles
84 Articles
Why ASEAN Must Prepare for AI Threats That Will Not Wait for Consensus
In July 2026, OpenAI models undergoing an internal cyber evaluation escaped the intended boundaries of the test and compromised parts of Hugging Face’s production infrastructure. The models chained vulnerabilities across both organizations to obtain answers from Hugging Face’s production database. OpenAI called it an “unprecedented cyber incident.” A narrow evaluation objective had driven an intrusion […] The post Why ASEAN Must Prepare for AI T…
AI Agents Go Rogue—Hugging Face Breached
The OpenAI–Hugging Face breach is the clearest signal yet that highly capable AI systems can operate as autonomous cyber actors—but it also shows that whether they become “too powerful to control” depends less on some mystical runaway intelligence and far more on how humans design, constrain, and govern them. Key Points During a security evaluation,...
Hugging Face CEO shares his demands of OpenAI after 'rogue' agent hack: 'It deserves an unprecedented response'
Clement Delangue, CEO of Hugging FaceHugging FaceHugging Face CEO Clem Delangue has some asks of OpenAI after an "unprecedented" security breach.He said he asked OpenAI to release all the traces from the "rogue" agent running on its models.He also asked OpenAI for $100 million in compute to bolster Hugging Face's cyber defenses.Even in the AI age, some conversations are best had in person.Last week, after Hugging Face suffered an unusual securit…
After the hacker attack on a technology company by an autonomous AI agent, the incident gets new explosiveness: According to insiders, the OpenAI agent was out of control for a longer time and once conspicuous than previously assumed.
According to a media report, developer OpenAI overlooked the outbreak and hacker attack of his AI for a long time. The company is said to have been warned that this could happen. Experts react horrified.
Coverage Details
Bias Distribution
- 39% of the sources are Center, 38% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
