The AI company Anthropic has issued a report on its efforts to disrupt so-called "threat actors" attempting to use its AI model Claude for malicious activities. Those include users doing research that could support the development of biological weapons.
The kind of science knowledge that experts say makes advanced AI useful in developing cures for diseases could also potentially be used to develop weapons. That risk is one of the big concerns that AI researchers and policymakers have about the technology.
Anthropic noted five case studies in its report. One involved researching how to make a mosquito-borne disease more transmissible. Another involved redesigning toxins for what Anthropic says was a national program.
Anthropic says it's hard to know definitively if the cases were related to plans to build bio-weapons, but notes that it disrupted the activity and shared the information with the authorities and partners in the industry.
Anthropic is a financial supporter of NPR.
An NPR editor adapted this audio report by John Ruwitch.
Copyright 2026 NPR
