AI and Pathogen Research: Balancing Scientific Opportunities and Biosafety Challenges
The use of AI in the life sciences has returned to the centre of debate following the publication of a report by Anthropic on applications of the Claude model in research involving viruses and toxins.
An article published in Science examines the controversy surrounding a report released by Anthropic, which identified five cases in which its Claude artificial intelligence platform was used to support research involving potentially sensitive biological agents. According to the company, there is no evidence that the researchers intended to cause harm, although some users reportedly attempted to bypass the platform’s access restrictions.
Three of the reported cases involved research on viruses, including studies related to chikungunya virus, avian influenza, and members of the orthopoxvirus family. According to the report, such work may contribute to the development of public health measures, vaccines, and disease surveillance systems, while also generating knowledge that could potentially be misused.
The report also describes two cases involving the study and computational redesign of toxins with potential therapeutic applications. These examples illustrate the dual-use nature of some life science research, a key issue within contemporary biosafety and biosecurity discussions.
Interpretations of the findings vary across the scientific community. Several researchers interviewed by Science argue that the activities described largely reflect standard practices in virology and biomedical research. Other experts, however, welcome Anthropic’s transparency and emphasize the need for continued development of oversight and governance mechanisms for AI systems used in the life sciences.
Beyond the specific controversy, the case highlights the growing importance of assessing the impact of artificial intelligence on biological research, particularly in areas with implications for biosafety and biosecurity. The ongoing debate underscores the need to balance scientific progress and technological innovation with the responsible management of emerging risks.
Source:
‘Chilling’ warning or overreaction? AI bioweapons report divides experts. Kaiser, J. (2026).Science. American Association for the Advancement of Science (AAAS).