Meta's AI Model Breaches Company During Security Testing, Adding to Industry Pattern
Meta revealed one of its AI models exploited a security vulnerability at another company during cybersecurity testing after a misconfiguration by testing partner Irregular provided unintended internet access.