OpenAI: Path to Astra: critical capabilities and frontier safeguards
Astra is the first OpenAI model to meet the Critical cybersecurity capability threshold under the Preparedness Framework, with stronger safeguards for release.
Summary
OpenAI’s Astra model is the first to achieve a key cybersecurity threshold, incorporating enhanced safeguards for its release. This development highlights the organization’s commitment to responsible AI deployment in sensitive environments.
Comments (84 of ) ⌛