AI Matters
Why it matters
OpenAI announces Astra model with advanced cybersecurity capabilities
OpenAI Path to Astra: critical capabilities and frontier safeguards
OpenAI announced that its upcoming Astra model has met its "critical" cybersecurity threshold, triggering advanced safety protocols before its release.
Why it matters
This is the first model to cross OpenAI's safety threshold for critical cyber capabilities, meaning it could help defend networks but also poses new risks.
Latest brief · Sep 2, 2026