A new report from Anthropic revealed five cases of “biological misuse” in which suspicious actors sought to use Claude AI to conduct deadly virus research and investigate toxic venom.

An Iran-linked actor used Anthropic’s Claude AI to collect publicly available information on U.S.

The report documents threat actors using Claude to research and develop drone swarms, anti-torpedo weapons systems, and targeting recommendations.

AI was used to develop missile guidance software in Yemen and to power cyber operations, a report alleges.

Anthropic's latest threat report reveals that its Claude AI model has already been misused to help target U.S. naval forces, build missile guidance software, hunt Uyghurs, and construct mass surveillance systems.

Anthropic said Alibaba, Moonshot, Zhipu, DeepSeek, and Xiaomi all tried to use Claude to train their models in various ways.

How will the watermarking actually work? Can it be hidden with editing? And how does this affect code?

Anthropic linked Alibaba to more than 151 million Claude exchanges and DeepSeek to another 12.1 million as Chinese labs sought advanced reasoning capabilities.

Anthropic said Thursday it has blocked efforts by bad actors to use its artificial intelligence models for malicious activity such as cyberattacks, surveillance, and research that could have led to biological weapons.