Anthropic Announces It Prevented Misuse of Artificial Intelligence Models
Anthropic has announced that it has successfully prevented the use of its artificial intelligence models for harmful purposes such as cyberattacks, surveillance, and biological weapon research.
Artificial intelligence company Anthropic announced that it has thwarted efforts by malicious actors to use its models for cyberattacks, surveillance activities, and research that could support the production of biological weapons.
Risks in Artificial Intelligence Models
With artificial intelligence models becoming more powerful, detailed cyberattacks no longer require advanced skills, and even individuals can create threats that were not possible in the past.
The company stated that it has implemented stronger security measures in its latest models to restrict biological research that could also be used in weapon manufacturing.
Detection of Malicious Activities
Between December 2025 and August 2026, Anthropic researchers detected numerous malicious uses, ranging from spyware vendors to politically motivated individuals and state-backed groups spreading propaganda.
It was determined that unnamed actors attempted to use the models for gain-of-function research aimed at increasing the transmissibility and immune evasion properties of the chikungunya virus.
Blocked Attempts and Measures
Anthropic reported that it successfully blocked a request asking for assistance from its Claude artificial intelligence model in writing a grant application for such research.
The company emphasized that all detected malicious activities were blocked and the information obtained was shared with official authorities and industry partners.