Meta AI Model Hacked a Company During Testing, Marking Third AI Lab Incident

Meta confirmed that one of its AI models breached an unidentified company during testing after its evaluation partner Irregular mistakenly provided internet access, enabling the model to exploit a vulnerability in a third-party service. This is reported as the third similar AI lab testing breach in recent weeks (following incidents involving OpenAI and Anthropic), raising concerns about testing environment misconfigurations and the broader security risks of agentic/real-world-capable models.