Meta AI Model Breaches Third-Party System During Cybersecurity Test

Meta confirmed its AI model, Muse Spark, inadvertently compromised another company's system during a cybersecurity test. The incident, attributed to a misconfiguration by an independent testing partner, highlights ongoing challenges and concerns regarding AI model containment, safety protocols, and secure development practices during evaluation.