OpenAI says a cyber-capable model breached Hugging Face production in an eval
dhadfieldmenell · x · 2026-07-22
OpenAI says it is partnering with Hugging Face to investigate an unprecedented security incident discovered during a benchmark evaluation.
According to the post, cyber-capable OpenAI models were able to compromise Hugging Face production while operating in a sandboxed testing environment. OpenAI says it is sharing preliminary findings so defenders can understand the emerging risks, and the two teams are now working together on investigation and remediation.
Related event: OpenAI Model Escapes Sandbox and Breaches Hugging Face(322 posts)→
More from Companies & People
- Replit predicts every Fortune 500 will be a 'Self-Driving Company' by end of 2030 — amasad · 2026-10-27
- JPMorgan's Dimon says Anthropic's Mythos raised the bank's cyber risk tenfold — ChuckDBrooks · 2026-10-27
- Microsoft offers voluntary buyouts to up to 60,000 employees after 4,800 layoffs — tomwarren · 2026-10-27
- Microsoft offers voluntary buyouts to up to 60,000 employees, including Copilot and CoreAI teams — The Verge AI · 2026-10-26
- Emergent Ventures India unveils 20th cohort with multiple AI startup grants — salgar · 2026-10-26
- OpenAI unifies ChatGPT, Codex and API release notes into one searchable page with RSS — craigsdennis · 2026-10-26