Anthropic says its models were misused for biological weapons research, surveillance and cyber attacks

Anthropic, an artificial intelligence company, announced on September 10, 2026 that its AI models were misused for harmful purposes, including the development of biological weapons, cyber attacks, influence operations, surveillance, and building weapons. The company released a detailed report on Thursday, detailing several instances where individuals attempted to misuse its Claude AI model for these malicious activities.

In response to recent high-profile incidents involving AI models, the company decided to share its threat intelligence report, which covered how they were able to disrupt every operation mentioned in the report.

One example cited in the report involved a user who entered a prompt into Claude tool seeking assistance with a grant application for researching the chikungunya virus. The report highlighted the potential risks associated with this misuse, as the chikungunya virus is a mosquito-borne virus that can cause debilitating symptoms and has no licensed therapeutic treatment.

The report did not disclose the institutional affiliation linked to the grant application, but Anthropic expressed concerns about the potential for this research to be less innocuous.

Anthropic stated that they have implemented several internal checks to address potentially problematic prompts and are actively working on improving these measures. The company also announced that it is banning accounts associated with the detected activities and is incorporating their investigative findings into new safety measures to prevent misuse of their services. The company emphasized the importance of balancing the utility of AI models for scientific research with the need to prevent their misuse.

Anthropic acknowledged the challenges in establishing appropriate safeguards, as AI models are increasingly being used in high-stakes applications and researchers warn of potential catastrophic consequences. The company believes that AI has the potential to significantly advance biological science and lead to the development of new medical treatments. However, they highlighted the need for careful regulation and oversight to ensure the responsible use of AI technology.

Public perception of companies at the forefront of AI development has become increasingly negative due to concerns about the safety and potential risks associated with these powerful models. Several legislators in the US Congress have introduced bills aiming to limit or halt AI research, although none of these proposals have yet become law.

Meanwhile, the AI companies are balancing the need for transparency and caution amid accusations of generating panic, while also attempting to project an image of responsible AI development.