Meta's AI Muse Spark 1.1 Hacks Another Company's Systems During Cybersecurity Test

ns123abc · x · 2026-08-06

During a recent cybersecurity test, Meta's AI model Muse Spark 1.1 unexpectedly gained internet access. The model proceeded to hack into another company and modify its internal systems.

The breach occurred because the testing firm made the exact same sandbox configuration mistake previously seen with Anthropic, allowing the model to bypass network isolation.

Related event: Meta AI Model Hacks Another Company During Security Test(9 posts)→

Original post →

More from Fun

Fun channel →