Loading...
Claude misused to develop biological weapons, Anthropic reveals
Anthropic's report also flagged five case studies

Claude misused to develop biological weapons, Anthropic reveals

Sep 11, 2026
01:12 pm

What's the story

Anthropic, a leading artificial intelligence (AI) firm, has revealed that its AI model, Claude, was misused for "malicious activity" supporting the development of biological and conventional weapons. The company's latest threat intelligence report highlights several cases of misuse over the past eight months. These included fake dating apps and hotel Wi-Fi scams as well as surveillance systems aimed at identifying dissidents.

Espionage incidents

Claude linked to cyber espionage campaign

The threat intelligence report also revealed that Claude was used in a Russia-linked cyber espionage campaign and by an Iranian propaganda institution.

These revelations come after warnings from a top safety researcher at Anthropic about AI's potential risks to humanity.

The company has also accused Chinese AI firms of attempting to replicate Claude's capabilities, further highlighting the global implications of its technology.

Mitigation measures

Anthropic disrupted 'malicious use' of its models

In light of these threats, Anthropic has taken steps to combat the misuse of its technology.

The company disrupted "Malicious use" of its Claude Haiku, Sonnet, and Opus models between December 2025 and August 2026.

However, it noted that none of the misuse cases involved Claude Fable or powerful Mythos-class models except for one instance of distillation, the process for training smaller AI models using larger ones.

ADVERTISEMENT

Security concerns

Biological misuse is one of the most serious risks

Anthropic's report also flagged "five case studies of actors using our models in ways that could support biological weapons development."

The company stressed that biological misuse is "one of the most serious risks of frontier AI models."

Without proper safeguards, such capabilities "could have catastrophic consequences," it added.

Jacob Klein, head of threat intelligence at Anthropic, said this situation is incredibly nuanced and not as straightforward as someone openly expressing intent to build a biological weapon.

ADVERTISEMENT

Weaponization risks

Claude also used to develop conventional weapons

The report also documented six cases where Claude was used "to develop software for conventional weapons, including firearms, missiles, armed drones," and more.

It noted that cybercriminals and state-backed hackers have increasingly leveraged its technology to aid their operations.

Among those named in the report were hacking group ShinyHunters and China-based labs.

Prevention strategies

AI advancing so rapidly it's a threat to humanity

In response to these threats, Anthropic has integrated its findings into its processes "to better prevent, detect, and disrupt these activities in the future."

The company has also shared intelligence with authorities and industry partners where appropriate.

This comes after a top safety researcher at Anthropic warned that AI is advancing so quickly he believes there is a greater than 10% chance it "could kill all humans" within the next decade.

ADVERTISEMENT