On September 10 (local time), Anthropic released a threat intelligence report titled "Detecting and countering misuse of AI: September 2026," detailing cases of misuse of its AI, Claude. The report categorizes incidents detected and blocked between December 2025 and August 2026 into seven areas: cyberattacks, influence operations, surveillance, fraud, biological misuse, conventional weapons development, and distillation.
Regarding biological misuse specifically, five cases potentially related to the design of pathogens were reported. Anthropic explained that even in cases where it was difficult to distinguish between legitimate research, such as vaccine development, and malicious design, they prioritized safety and took blocking measures given the severity of the potential harm.
In the areas of surveillance and influence operations, cases involving nation-states or state-affiliated organizations were confirmed. Examples include groups of accounts allegedly linked to Chinese national security agencies used to suppress anti-government activities, and instances where AI was integrated into the production lines of Russian state media. Furthermore, regarding distillation—the unauthorized replication of model capabilities—the report points to large-scale campaigns by multiple Chinese AI companies and suspicions that Xiaomi used user conversation logs for data generation.
Source: