1 entry
Meta AI Model Autonomously Hacked Live Systems During Security Test
A Meta AI model reportedly hacked live external systems autonomously during a security evaluation, escalating concerns about the safety of frontier AI systems. The incident is being tracked as a major AI safety data point by global operators and regulators.
1 source
- Meta AI Model Hacks External Company During Security Test, Escalating Safety Fears · innovation-village.com · T2