Developers have identified a number of cases in which AI could potentially be used to develop biological weapons
The American company Anthropic reported five instances in which its artificial intelligence model, Claude, may have been used in research potentially related to the development of biological weapons.
The company emphasized that it does not know the exact plans of the researchers who used Claude, nor is it disclosing the names of the relevant research organizations or countries.
According to Anthropic, the problem is that the same biological methods can be used both to create medicines and vaccines and to develop dangerous pathogens.
“You don’t see someone saying, like in a comic book, ‘Hey, I want to create a biological weapon and kill everyone,’” said Jacob Klein, head of threat analysis at Anthropic.
One of the identified cases occurred in May 2026.
According to Anthropic, a scientist asked Claude to help prepare a grant proposal to fund research on the chikungunya virus.
According to the project description, the research involved genetically modifying the virus, studying how it spreads, and examining its ability to evade the body’s immune response.
The work was planned to be conducted at a military research institute.
Anthropic notes that such cases demonstrate the complexity of controlling the use of modern artificial intelligence models.
Scientific inquiries, which may appear legitimate on their own, could potentially be used to create or improve dangerous biological agents.
The company does not claim that all five research projects were specifically aimed at creating biological weapons. These are cases that raised suspicions due to the nature of the requests and the dual-use potential of the relevant technologies.
This was reported by the company’s press office.
In Dublin, Ireland, European Commission Executive Vice President Henna Wirkkunen spoke about the need for global rules to regulate artificial intelligence. She spoke on the sidelines of a conference on child safety online following a warning from a former Anthropic engineer that AI could pose a threat to humanity.
In Geneva, Switzerland, UN High Commissioner for Human Rights Volker Türk stated that the development of advanced artificial intelligence systems could pose an existential risk to humanity. He called for the urgent implementation of robust safety safeguards for such systems.
