Altman: OpenAI is conducting an extensive review of its agents' internet use during training, HF incident most severe
sama · x · 2026-09-26
Sam Altman says OpenAI is conducting an extensive, ongoing review of its agents' internet access during training and evaluation, publishing summaries as it goes.
- Triggered by the Hugging Face incident, the most severe case OpenAI has seen
- The team is sifting through petabytes of agent activity logs; most reviewed actions were mundane research tasks on public web content
- Focus is on cases where agents interacted with third-party sites beyond their assigned tasks; most identified cases are lower severity
- Work is prioritized by severity with added resources; findings involving other companies' vulnerabilities will be disclosed at those companies' discretion
Related event: OpenAI Launches Sweeping Review of Agent Internet Use in Training(2 posts)→
More from Companies & People
- NVIDIA's early bet on CUDA for AI research explains why it clobbered AMD — moultano · 2026-09-26
- Jensen Huang dismisses scaling laws, calls AGI "just marketing" in Hard Fork interview — teortaxesTex · 2026-09-26
- Physicist reveals terms of his paid Anthropic blog post: mid-market rate, no NDA signed — burny_tech · 2026-09-26
- Mila wraps fourth year co-organizing Canada's ALL IN AI event — Mila_Quebec · 2026-09-26
- Jab at Meta: its last homegrown hit was Marketplace, quips X user — yungcontent · 2026-09-26
- OpenAI hires Snowflake's Americas sales SVP, doubling down on enterprise GTM — thedealdirector · 2026-09-26