Muse Spark 1.1

Meta AI Model Breaches Third-Party Service During Cybersecurity Assessment

A configuration error in a testing environment allowed a Meta AI model to gain internet access and exploit a vulnerability in an external system.