Anthropic reports five cases of AI misuse linked to biological weapons risk

Anthropic says it identified and disrupted five cases in which scientists used its Claude artificial intelligence models in ways that could support biological weapons development.

The cases were detailed in the company’s threat intelligence report, covering alleged misuse between December 2025 and August 2026. Anthropic said the activity did not establish that a biological weapon had been developed or that an attack was planned.

The company described biological misuse as one of the most serious risks associated with frontier AI, warning that such capabilities could have catastrophic consequences without appropriate safeguards. It also noted that the same information could potentially be used for beneficial purposes, including developing vaccines or treatments.

Anthropic’s report identified six additional cases involving conventional weapons, including firearms, missiles, armed drones and bombs, as well as targeting and control systems. It also described alleged misuse connected to cyber operations, surveillance, scams, fraud and influence campaigns.

The report said Claude had been used by actors linked to a Russia-based cyber espionage campaign and an Iranian propaganda institution. Anthropic said it had incorporated its findings into measures to prevent, detect and disrupt misuse, and shared intelligence with authorities and industry partners where appropriate.

The disclosures add to wider calls for controls on advanced AI. Google recently said someone attempted to use its Gemini system to obtain instructions for synthesising weaponised biological agents, while US Senator Bernie Sanders has called for a pause on advanced AI development and a ban on artificial superintelligence.