Anthropic announced Thursday, September 10, that it has disrupted several attempts by unethical entities to misuse its artificial intelligence (AI) models for malicious activities, including research that can lead to the development of biological weapons.
Releasing its third threat intelligence report, Anthropic revealed it had blocked efforts made between December 2025 and August 2026 involving spyware vendors, state-sponsored groups, and politically motivated individuals.
The report also added snippets of malicious code and AI prompts that were detected and stopped. One of the cases was a prompt that wanted Claude’s assistance in writing a grant application for gain-of-function research on the chikungunya virus, a mosquito-borne illness causing severe pain and fever. The research was focused on improving the virus’s transmission and immune evasion properties. Anthropic noted that it can be used for legitimate vaccine development but also make the pathogen more dangerous.
Anthropic’s head of threat intelligence, Jacob Klein, said: “You are not seeing someone in a comic book kind of way, ‘Hey, I want to build a biological weapon to kill everybody. It’s an incredibly nuanced situation.”
Claude was also reportedly used for developing conventional weaponry like firearms, missiles, weaponised drones, and explosives. In addition to this, the report stated that Russian state media had employed the AI technology to produce propaganda, such as a fictitious story about Moldova’s election results, while cyber espionage and surveillance were carried out by actors associated with Russia and Iran.
According to Anthropic, older versions of the AI were “well below the threshold” for use in dangerous biological experiments, but admitted that the new, more advanced AI required enhanced security measures. Restrictions have been placed on dual-use biological queries in the latest version, namely Claude Fable 5.
The announcement came amid increased calls for regulation of AI technology. Bernie Sanders, a US senator, has recently filed a bill to ban AI superintelligence.
2026-09-11 21:47:00
