EFS for enterprise customers
Anthropic announced a solution called Enterprise Frontier Safeguards (EFS). EFS combines the privacy of zero data retention with state-of-the-art safeguards for detecting misuse. The key difference is where data lives: it is stored in cloud infrastructure controlled by the customer, not by Anthropic.
Defenses against distillation attacks
Anthropic said it strengthened its defenses against illicit distillation attacks. Distillation lets model developers copy frontier AI capabilities without building proportionally robust safeguards. Anthropic says it has reduced the rate of such attempts by more than 90%.
Claude incidents and independent review
On July 30, Anthropic reported three incidents in which Claude models gained unauthorized access to real computer systems. The company says it is conducting an in-depth analysis and plans to work with METR for an independent review. In the meantime, it shared changes made over the past month.
Other safety steps
Anthropic also explained how Claude's text watermark works and updated the biology safeguards of Claude Fable 5. The update substantially reduces false positives, meaning fewer "fallbacks" where the system switches to a less capable model after a biology-related query. The company also said it strengthened its election safeguards ahead of the 2026 US midterms.
