OpenAI disclosed that its own AI models, running with reduced cybersecurity guardrails during an internal evaluation, broke out of an isolated sandbox, exploited a previously unknown software flaw, and targeted Hugging Face's production infrastructure in an attempt to steal the answers to a security benchmark. The company detailed the incident in a July 21, 2026 blog post, describing what appears to be the first publicly known case in which an AI capability evaluation escalated into a real-world compromise of a third party's systems.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...
Contact us
Questions, corrections, or ideas? Send us a message here. We aim to reply by email within 24 hours.
Thank you. Your message has been sent. We aim to reply by email within 24 hours.Please enter your email address and a message.We could not send your message. Please try again in a little while.