Anthropic, creator of the artificial intelligence assistant ‘Claude’, has already prevented at least five research projects potentially related to the development of biological weapons, some with high pandemic potential this year.
© Getty Images
11/09/2026
by
Lusa
Segund report “Detect and combat the misuse of AI”, released on Thursday, at least four cases were linked to state organizations that Anthropic decided not to disclose, given the possibility of being only academic studies.
The first of the examples focused on research and development overactivity to make the chikunguna virus more transmissible and resistant to the immune system.
The company highlighted that the knowledge obtained could be used in the creation of vaccine and treatments, although the fact that the grant form indicates that scientists would be civilians, but the investigation would take place in a military laboratory, has led to the blocking of the ‘Claude’ account.
According to the World Health Organization, chikunguna is a disease transmitted by mosquitoes that causes fever and “incapacitating” pain in joints.
Anthropic found the intention of a state body to use its AI models to study non-contagious poisons. And he ended up blocking another user who was investigating bacterial toxins and proteins from viral hemorrhagic fevers with pandemic capacity.
The American technology has also detected similar initiatives, but with less unsuspecting purposes, to investigate the smallpox-causing virus or the passage of avian influenza to mammals.