Anthropic Report Highlights Growing AI Misuse, Warns of Biological Weapons Risk

Anthropic says it has blocked attempts to exploit its AI models for malicious activities that could potentially contribute to the development of biological weapons. The AI developer published the findings in its recent threat intelligence report, adding to the growing list of warnings from industry practitioners and researchers about the technology’s potential risk to humanity. 

Anthropic Acknowledged Biological Misuse Risk

Anthropic’s threat intelligence report showed that malicious actors had used Claude AI in ways that could lead to the development of biological weapons at least five times. The AI firm acknowledged that biological misuse was one of the “most serious risks of frontier AI models” that could have catastrophic consequences in the absence of correct safeguards. 

"The same information that can be used to develop a biological weapon could also be used to develop, for example, a vaccine or a cure for a disease," Anthropic noted in the report, as per  BBC.

Anthropic also reported instances where Claude AI was used to build software “for conventional weapons, including firearms, missiles, armed drones, bombs, and other munitions, as well as the targeting and control systems that operate them".

Malicious Actors Used Claude AI for Cyberattacks

Anthropic’s AI misuse report showed that malicious actors had used Claude AI models for cyberattacks, surveillance, and propaganda. In one instance, a hacking group whose operations resemble those of Russia-based Midnight Blizzard used Claude AI. 

The cybercriminal group had developed a system that automatically detects when security defences detect its malware and rewrites code until it evades detection. The report also named ShinyHunters and China-based labs among the cybercriminals that have used Anthropic AI models. 

Anthropic also reported that increasingly, cybercriminals and state-backed hackers were using Claude models in their operations. The AI company said actors linked to a Russia-based cyber espionage campaign and an Iranian propaganda institution had used its models. 

Which Claude AI Models Were Misused? 

Based on Anthropic’s report that malicious actors appear to have misused Claude Sonnet, Haiku, and Opus models. The company reported that it disrupted malicious activities in these models between December 2025 and August 2026. 

The AI developer noted that none of the AI misuse cases involved Claude Fable or Mythos-class models, except for one distillation case. Anthropic also accused Chinese AI companies of attempting to replicate Claude AI’s capabilities. 

The AI misuse report comes days after an Anthropic researcher, Jacob Coxon resigned. Coxon warned that top AI companies are pursuing increasingly capable systems without adequate safeguards. 

What Anthropic’s Report Means for the AI Industry

Anthropic’s latest report underscores the growing challenge facing AI developers and the industry at large. As AI models become more powerful and capable, their abilities can also be exploited for harmful purposes. The company called for stronger safeguards, greater transparency, and closer cooperation between AI firms and governments in addressing emerging threats in the industry.