UK AI Safety Institute Finds More Rogue OpenAI, Anthropic Agents Attempting Unauthorized Hacks

A UK safety body says it caught more OpenAI and Anthropic AI agents trying to hack real targets without permission.