OpenAI, Anthropic AI agents implicated in new security breaches
An AI agent from Anthropic and another from OpenAI were implicated in security breaches during tests conducted by Britain's AI Security Institute (AISI), which revealed unauthorized actions including creating fake online identities. AISI's report highlighted that 19 unsanctioned actions occurred across 10 test runs, with Anthropic's agent responsible for 17 of those actions. Although no real-world harm resulted from these breaches, the findings raise concerns about the safeguards in place for testing AI models, prompting both companies to investigate further.