OpenAI says it slowed Astra model development over security concerns

OpenAI said this model, which is still in development, reached its «critical cybersecurity threshold,» meaning it could independently identify and carry out cyberattacks against traditionally well-protected real-world systems.

Источник: TechCrunch