Anthropic Announces Prevention of AI Model Misuse
The company announced that it has blocked attempts to use artificial intelligence models for harmful purposes such as cyberattacks and biological weapon research.
Anthropic has announced that it successfully blocked attempts by malicious actors to use artificial intelligence models for activities such as cyberattacks and biological weapons research.
AI Safety and Blocked Attempts
Anthropic announced that it has successfully blocked attempts by malicious actors to use artificial intelligence models for cyberattacks, surveillance activities, and research that could produce biological weapons. As AI models become more powerful, sophisticated cyberattacks no longer require specialized expertise, allowing even individuals to generate threats that were not possible in the past.
Detected Cases of Misuse
Between December 2025 and August 2026, company researchers detected activities by spyware vendors, politically motivated individuals, and state-backed disinformation groups. It was noted that some unnamed actors requested support for grant applications regarding gain-of-function research on the chikungunya viruse aimed at increasing immune evasion and transmissibility, and this request was blocked by the systems.
Future Precautions and Industry Cooperation
This report was published following the resignation of a researcher who drew attention to safety concerns in AI development processes. Anthropic emphasized that older models have fewer restrictions against advanced biological research, but newer models feature robust safeguards.