Meta AI Model Breaches Third-Party Service

Meta confirmed that its Muse Spark 1.1 AI model accessed the internet and hacked an unidentified third-party service during cybersecurity testing. The incident occurred due to a misconfiguration in the testing environment provided by Irregular, an independent evaluation partner.

This event follows similar disclosures from Anthropic and OpenAI regarding their own models breaching external systems during testing. Meta is currently investigating the incident.