‘Malicious Misuse’: Anthropic Says It Blocked Scientists Who Used AI For Bioweapons Research

Last Updated:September 11, 2026, 02:40 IST

Anthropic said the cases were among the “most notable and novel threat activity” it had identified so far.

The AI company detailed five biological misuse cases in a new threat intelligence report. (AI generated image)

Anthropic on Thursday said it had blocked scientists who used its Claude AI models in ways that could support biological weapons development, highlighting growing concerns over how advanced artificial intelligence could be misused for dangerous research.

The AI company detailed five biological misuse cases in a new threat intelligence report covering activity it detected and disrupted between December 2025 and August 2026.

The report also described harmful uses involving cyber operations, surveillance, scams, conventional weapons development and propaganda.

Anthropic said the cases were among the “most notable and novel threat activity” it had identified so far.

Five Cases Raised Bioweapons Concerns

Among the cases were efforts to research how a mosquito-borne disease could be made more transmissible and work involving the redesign of toxins as part of what Anthropic described as a national programme.

The company said it was difficult to establish whether the users actually intended to develop biological weapons. It stressed that those involved were working scientists and that it was not claiming they intended to cause harm.

Anthropic also withheld the names of the scientists, their institutions and countries, as well as details about the biological agents and research techniques involved.

“When we detected and investigated these cases, we banned the users’ accounts,” the company said, adding that its findings were used to strengthen model safeguards, enforcement systems and threat intelligence.

AI’s Dual-Use Biology Risk

Anthropic described biological misuse as “one of the most serious risks” posed by frontier AI models.

The company said the cases showed that its models had capabilities that could potentially support dangerous biological work, but stressed that they did not prove those capabilities would be used to create biological weapons in the real world.

“The individuals implicated in these case studies are working scientists. We do not assert that they intended harm, and identifying them or their labs could expose them to harm,” the report said.

AI-assisted biological research has a clear dual-use problem, Anthropic said. The same information that could potentially help develop a biological weapon could also contribute to medical research, including vaccines or treatments for disease.

The company said newer models have stronger safeguards and restrict access to a wide range of dual-use biological research queries.

Anthropic Tracks Wider Threats

The biological cases formed only part of Anthropic’s latest report, which examined malicious activity across seven areas.

The company said its threat intelligence team had disrupted operations involving suspected state-backed groups, financially motivated criminals, commercial spyware firms, propaganda institutions and politically motivated individuals.

Other cases included surveillance systems designed to identify dissidents and networks of fake dating apps allegedly created to defraud users.

Anthropic said it had shared relevant intelligence with authorities and industry partners and would continue improving its safeguards as threat actors try to bypass them.

The company said the findings were intended to help other AI developers recognise similar patterns and “give governments and civil society” a clearer picture of emerging AI-enabled threats.

Quick Answers

Powered by

Ask News18

What did Anthropic’s threat intelligence report reveal about scientists misusing Claude for biological research?

Anthropic revealed it blocked working scientists who used its Claude AI models in ways that could support biological weapons development. The company detailed five biological misuse cases detected between December 2025 and August 2026. These cases included researching how to make a mosquito-borne disease more transmissible and redesigning toxins as part of a national programme. Anthropic banned the users’ accounts but did not release their names, institutions, or countries to protect them from potential harm.

Why is AI-assisted biological research considered a dual-use risk by Anthropic?

Anthropic explains that AI-assisted biological research has a clear dual-use problem because the same information and techniques used to develop biological weapons can also contribute to legitimate medical research, such as creating vaccines or treatments for diseases. Because of this overlap, Anthropic must walk a difficult line to determine if research is legitimate or nefarious, opting for a restrained approach with prompts because the consequences of missing malicious activity are extremely severe.

What other security threats did Anthropic disrupt besides biological weapons research?

Anthropic’s threat intelligence report detailed disruptions across seven areas, including cyber operations, surveillance, scams, conventional weapons development, and propaganda. Specific cases involved suspected state-backed groups, financially motivated criminals, commercial spyware firms, and politically motivated individuals. These included Chinese and Iranian government-linked actors targeting dissidents for surveillance, networks of fake dating apps used to defraud users, and Russian media outlets generating online propaganda.

Powered by

Ask News18

About the Author

News Desk

The News Desk is a team of passionate editors and writers who break and analyse the most important events unfolding in India and abroad. From live updates to exclusive reports to in-depth explainers, …Read More

Read More