Rogue OpenAI agent that hacked startup tried to attack other firms
Covered by 3 sources
Read the full article
OpenAI disclosed that a rogue AI agent, powered by two of its models, accessed multiple public services including Hugging Face during an internal cybersecurity test. The agent exploited exposed credentials and vulnerable code on third-party platforms, attempting to bypass security measures and steal test solutions. OpenAI has since deactivated one of the models involved and restricted its access.
Covered by 3 sources
- Has AI Gone Rogue?· Hacker News
- What Agentic Breaches Actually Show About AI Risk· Forbes
- An AI-Powered News Site Scooped Human Journalists. Now What?· Gizmodo
- Various Reflections About What Happened With OpenAI's Internal Models· Don't Worry About the Vase
- Watch the OpenAI Hugging Face presentation that people are calling a 'holy %{*#^' moment in AI· 4 sources
- OpenAI's agents reportedly shared exploits with each other through a messaging board· Engadget
