Anthropic blocks effort to use AI to research potential biological weapons

Anthropic says it has blocked efforts to use its artificial intelligence models to carry out cyber attacks and research which could have led to the creation of biological weapons.

The US company has published a lengthy report into AI misuse, including instances of malicious code and dangerous prompts it has discovered in its models.

Between December 2025 and August 2026, Anthropic's researchers identified misuse by "politically motivated individuals, state-sponsored groups seeking to spread propaganda, and spyware companies".

It blocked one request for Claude to help create a grant application for scientific funding into research to "create a new or enhanced biological property" of the chikungunya virus.

The mosquito-borne virus triggers symptoms such as severe pain and fever, and the request sought to enhance mutations to make the virus more harmful.

Anthropic said it could have been "used to make the pathogen more dangerous".

None of the cases included in the report used the newer Claude Fable or Mythos-class models, except for one "industrial-scale, covert campaign to extract a model's capabilities and replicate them in another model without authorization".

Anthropic said the older models, such as Claude Opus 4 and Claude Sonnet 4.5, "were well below the threshold" to assist a user in carrying out dangerous biological research.

The report said: "As a result, safeguards on these models were less stringent, directed mostly at preventing access to content that might uplift novices in recreating known bioweapons.

"But for today's models - which are capable of assisting in a range of complex scientific research tasks - the evidence is no longer certain, and we cannot make that same assurance."

Anthropic said it has implemented "stronger safeguards that restrict access to a wide range of dual-use biological research queries".

Read more from Sky News:
Timeline of 9/11
US voters warned it's 'America or Trump'

Experts have repeatedly called on governments to regulate AI themselves as the models become more sophisticated, rather than relying on the industry itself.

Earlier this week, one of Anthropic's researchers, Jacob Coxon, announced he was resigning and said the company, along with its competitor OpenAI, "are racing straight to self-improving superintelligence and gambling with our lives".

He also warned that some of his colleagues believe AI could threaten humanity by the end of the decade.

(c) Sky News 2026: Anthropic blocks effort to use AI to research potential biological weapons