Anthropic has blocked multiple accounts after detecting attempts to use its artificial intelligence models for research that could potentially support the development of biological weapons.
The company behind Claude identified biological misuse as one of the most serious risks linked to increasingly capable AI systems in a report published Thursday.
Anthropic reviewed 30 days of activity and identified about 35 distinct research efforts involving potentially concerning biological work. However, the company said it could not determine whether the users intended to cause harm or were conducting legitimate scientific research.
The report detailed five cases involving researchers who allegedly tried to bypass Anthropic’s safeguards. Some reportedly accessed the service from restricted regions or disguised the purpose of their research.
The cases involved potentially dangerous gain-of-function research, infectious diseases and studies involving novel venoms and toxins. Researchers also worked on subjects including bird flu and orthopoxviruses.
Manchester United Return to Champions League With 4-0 Win Over Sabah
In one case, a user sought help preparing a grant application for gain-of-function research on the transmissibility and immune evasion of chikungunya virus.
Another researcher outside the United States used Claude while working on bird flu. The research included questions about how viruses adapt to mammals and cause severe disease.
Anthropic said the individuals were working scientists but did not disclose their institutions or countries.
The company warned that increasingly capable AI systems could lower the expertise and resources traditionally needed to conduct dangerous research. It described this as a major challenge for AI safety.
At the same time, AI-assisted drug discovery remains one of the technology’s most promising applications. Researchers are exploring AI to accelerate the development of treatments for serious diseases.
Anthropic warned that the same capabilities could also be exploited for harmful biological research. Its latest models, including Claude Fable 5, feature stronger safeguards designed to restrict access to a broad range of dual-use biological requests.
The company also said it had disrupted attempts to use its systems for surveillance operations, scams and the development of conventional weapons, including drones and missiles.
The disclosure comes as AI researchers raise concerns about the rapid development of increasingly powerful systems. Former Anthropic researcher Jacob Coxon recently warned that future AI systems could potentially be used against critical infrastructure or to create highly destructive biological weapons.
In July, nearly 1,400 AI industry employees signed an open letter calling for stronger government regulation and a slower pace of AI development. OpenAI Chief Scientist Jakub Pachocki has also warned that AI capabilities are advancing faster than researchers’ ability to reliably monitor and control them.
Anthropic said the AI industry needs a lawful and verifiable way to cooperate on controlling the release of increasingly powerful models. The company argued that stronger safeguards and coordinated oversight will be necessary as AI capabilities continue to advance.
