Anthropic opens its most powerful AI models to more security teams
Anthropic is expanding a program that allows vetted cybersecurity professionals to test its most powerful AI models with fewer safeguards, after its Project Glasswing initiative
October 07, 2026