Countering misuse of AI: September 2026 / Anthropic \ Anthropic
www.anthropic.com
Oct. 3, 2026, 4:03 a.m.
Anthropic's Threat Intelligence team has released a comprehensive report documenting the detection and disruption of malicious Claude AI usage between December 2025 and August 2026. The report reveals that threat actors—including state-sponsored groups, financially motivated criminals, commercial spyware vendors, and politically motivated individuals—attempted to exploit Claude's Haiku, Sonnet, and Opus models across seven harm categories: cyber operations, influence operations, surveillance, scams and fraud, biological misuse, conventional weapons development, and distillation. Notable cases included fake dating app networks designed for fraud and surveillance systems targeting dissidents. No misuse involved the Fable or Mythos-class models, except for one distillation case. Each incident was disrupted, and learnings were incorporated into strengthened safeguards while intelligence was shared with authorities and industry partners. Anthropic emphasizes that as AI capabilities advance, proactive disclosure and cross-industry coordination become essential for identifying threat patterns, informing policy responses, and collectively strengthening defenses against emerging malicious applications.