OpenAI's Astra Model Labeled Critical Cyber Risk as Release Nears
OpenAI's Astra model is the first AI system rated a critical cybersecurity risk, prompting limited access and growing concern among researchers as its public release approaches.
Read the full story → · Latest coverage 3 Sept 2026, 7:13 pm UTC
Key moments · 7
OpenAI announces plans to limit Astra's advanced cyber tools
Technology ↗Astra rated 'critical' cyber risk after internal testing
The Wall Street Journal Tech ↗Astra becomes first AI model at critical cybersecurity threshold
SecurityWeek ↗OpenAI says Astra needs more safety before launch
The Hill ↗Researchers warn of safety disaster ahead of Astra release
The Verge - All Posts Tech ↗OpenAI begins rolling out Astra model to select users
CNBC ↗OpenAI launches GPT-6 Astra amid scrutiny over agent safety
World News CNA ↗Coverage by day
newest firstLoading coverage…