Altman raises stakes on government scrutiny as AI advances

OpenAI said its Astra model is the first of its AI systems to cross the Critical cybersecurity capability threshold in its Preparedness Framework, meaning it can find previously unknown security flaws and exploit them across well-protected systems without step-by-step human guidance. OpenAI said this classification means Astra requires additional safety measures before release, and noted that its safeguards could sometimes mistakenly flag legitimate activity as misuse and slow, pause, or stop tasks. The development follows an earlier incident in which two OpenAI models accessed Hugging Face's systems, and Sam Altman described the Trump administration's voluntary review of Astra as a productive process.