Anthropic identified and blocked five cases in which actors used its Claude models in ways that could support biological weapons development. The activity occurred between December 2025 and August 2026. The firm published the findings in its threat intelligence report. The cases form part of a wider pattern of misuse that also included cyber espionage, influence operations and conventional weapons software.
What did Anthropic report about biological weapons risks?
Anthropic stated that its models were used in five separate instances that could assist biological weapons development. The company described biological misuse as one of the most serious risks of frontier AI models. It noted that the same information useful for weapons could also advance vaccines or disease treatments, creating a narrow distinction that requires careful safeguards. The report covered activity detected across Claude Haiku, Sonnet and Opus models. None of the biological cases involved the newer Claude Fable or Mythos-class models, with the exception of one instance of distillation. Jacob Klein, the head of threat intelligence at Anthropic, told the New York Times it was an incredibly nuanced situation because users rarely state explicit harmful intent in direct terms. Anthropic warned that without the correct safeguards such capabilities could have catastrophic consequences. The report highlighted five case studies of actors using the models in ways that could support biological weapons development. Biological misuse is described as one of the most serious risks of frontier AI models.
Which other misuse cases appeared in the report?
The report documented misuse for cyber operations, influence campaigns, surveillance, scams and conventional weapons development. Six cases involved actors using the models to create software for firearms, missiles, armed drones, bombs and targeting systems. The cases of concern detected over the past eight months ranged from fake dating apps and hotel Wifi scams, to surveillance built to identify dissidents. The report also noted that a hacking group whose work is consistent with the Russia-based Midnight Blizzard allegedly used AI to build a system that automatically detected when its malware was flagged by security defences and rewrote code until it evaded detection. Anthropic has also accused Chinese AI firms of trying to replicate Claude's capabilities. The lengthy report lists cases in which suspected state-sponsored groups, criminals, spyware vendors, state propaganda institutions and politically motivated individuals misused its technology. Claude was also used by actors linked to a Russia-based cyber espionage campaign and by an Iranian propaganda institution.
Who were the main actors named by Anthropic?
Suspected state-sponsored groups, criminals, spyware vendors, state propaganda institutions and politically motivated individuals misused its technology. Claude was also used by actors linked to a Russia-based cyber espionage campaign and by an Iranian propaganda institution. Hacking group ShinyHunters, as well as China-based labs, were among those named in the report. The lengthy report lists cases in which these groups misused Claude Haiku, Sonnet, and Opus models. Anthropic said it had shared intelligence with authorities and industry partners where appropriate. The cases ranged from fake dating apps and hotel Wifi scams to surveillance built to identify dissidents and influence operations.
How have policymakers and researchers responded to these risks?
US Senator Bernie Sanders called for a temporary pause on advanced AI development and a ban on artificial superintelligence. He stated on BBC Newsnight that a chance of cataclysmic impact required immediate slowdown measures. OpenAI chief scientist Jakub Pachocki urged voluntary slowdowns until safeguards are in place. A top safety researcher at Anthropic warned AI is advancing so quickly he believes there is a greater than 10% chance it could kill all humans within the next decade. President Donald Trump rejected calls for pauses, saying on Thursday he was concerned if we don't win AI, we're going to be put in a very bad position. The warning prompted an open letter to UK Prime Minister Andy Burnham calling for a new multinational treaty for the safe development of AI.
What actions has Anthropic taken following the findings?
Anthropic incorporated the report's conclusions into its detection and prevention processes to better prevent, detect, and disrupt these activities in the future. The company said it would continue to monitor and disrupt malicious activity. The report is the company's first this year and follows regular industry practice also followed by Google, which reported a similar attempt to obtain instructions for synthesising weaponised biological agents. The California-based company said it had incorporated its findings into its processes. Anthropic has shared intelligence with authorities and industry partners where appropriate.
Frequently asked questions
What specific biological misuse did Anthropic detect?
The report listed five case studies where actors used Claude models in ways that could support biological weapons development, though exact technical details were not disclosed.
Did any misuse involve the most advanced Claude models?
No biological misuse cases involved Claude Fable or Mythos-class models, with one exception of model distillation.
Which countries were linked to the actors in the report?
Actors consistent with Russia-based and Iran-based operations were named, along with China-based labs and hacking groups.
Has Anthropic shared its findings with governments?
Yes, the company stated that it shared intelligence with authorities and industry partners where appropriate.
What is the current US policy position on AI pauses?
President Trump has rejected calls for pauses, while Senator Sanders has introduced legislation to pause advanced development and ban superintelligence.
Key takeaways
- Anthropic blocked five cases of potential biological weapons support between December 2025 and August 2026.
- Six additional cases involved conventional weapons software development.
- Russian, Iranian and Chinese actors were among those named in the threat report.
- Senator Sanders called for a pause on advanced AI and a ban on superintelligence.
- Anthropic has updated its safeguards and shared findings with authorities.
Outlook for AI safety governance
The Anthropic report adds to a series of industry disclosures that highlight both the dual-use nature of frontier models and the difficulty of distinguishing legitimate from harmful queries. Lawmakers in the United States and the United Kingdom have begun discussing treaties and slowdown mechanisms, while companies continue to refine detection systems. The coming months will show whether voluntary measures or formal regulation gain broader support.