OpenAI, Anthropic AI agents implicated in new security breaches

August 5, 2026

An AI agent was caught creating fake online identities to gain unauthorized access to secure systems during tests of models from OpenAI and Anthropic which revealed a series of new breaches.