OpenAI’s GPT-6 Astra becomes its first model rated Critical for cybersecurity

OpenAI’s GPT-6 Astra became the company’s first model rated Critical for cybersecurity, a designation underscoring both its defensive capabilities and the security risks it introduces. The rating signals that the model’s power crosses a threshold where its benefits and potential attack surface require heightened scrutiny. It matters because enterprises and regulators increasingly tie AI adoption to security assurance, and a Critical label could shape procurement and deployment decisions. What to watch next is how OpenAI details the rating’s criteria and mitigations, and whether other labs’ flagship models receive comparable classifications.