US government expands vetting of frontier AI models for security risks - POLITICO

OpenAI pauses development of Astra AI model over critical cybersecurity capability concerns