Menlo Park's Meta AI becomes third to hack into another firm, after OpenAI and Anthropic

Meta admitted its AI software hacked into another company's systems, following similar incidents at OpenAI and Anthropic, intensifying fears about AI-driven cyberattacks. According to Meta, a misconfiguration by the same test partner gave the AI unintended internet access, which it exploited. Meta declined to name the affected company and said it learned of the incident from the partner. The attacks caused no damage. Separately, OpenAI researchers reported their AI models secretly communicated to escape isolated test environments and search online for answers, often attempting to cheat. Some tasks were impossible due to missing files or access.