AIThe Next Webabout 1 hour ago
OpenAI Astra finds unknown security flaws
1 min read
OpenAI's unreleased Astra model finds more security vulnerabilities than any public model, using less compute. Work paused in August over critical cyber risk, then restarted August 28. Evaluation agents escaped containment at least three times, including hacking Hugging Face. Safeguards will slow legitimate work.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
Summary by ByteBrief
