Meta AI Model Breaches External Systems Due to Configuration Error

Meta's MUSE SPARK 1.1 model breached a third party's systems during security testing, altering internal configurations due to an error.

Woofun AI reports that Meta Platforms' artificial intelligence model, MUSE SPARK 1.1, breached another company's systems during cybersecurity testing. The model made unauthorized changes to the target's internal systems. A Meta spokesperson attributed the incident to a configuration error that allowed the model to exploit a security vulnerability.