Anthropic Releases Report Saying It Disrupted Attempts to Use AI for Biological Weapons and Cyberattacks

Artificial intelligence startup Anthropic has released a threat intelligence report revealing that it identified and disrupted multiple attempts to use its Claude models for malicious activities, including dual-use research that could assist in biological weapons development, cyber operations, and automated propaganda campaigns.

Covering incidents detected between December 2025 and August 2026, the report highlights five case studies where actors attempted to leverage AI capabilities to support biological research. In one instance, systems intercepted a request asking the model to assist in drafting a grant proposal aimed at mutating the chikungunya virus to render it progressively more dangerous.

While acknowledging that dual-use scientific information can assist legitimate medical breakthroughs, Anthropic emphasized that without strict controls, such queries could pose severe risks:

"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 said.

Addressing the complexity of detecting harmful intent, Jacob Klein, head of threat intelligence at Anthropic, said that it is "an incredibly nuanced situation".

"You are not seeing someone in a comic book kind of way say, 'Hey, I want to build a biological weapon to kill everybody,'" he said.

Beyond biological concerns, the report detailed six instances involving conventional weapons development such as missile control software and drone systems alongside cyber operations linked to groups like Midnight Blizzard and ShinyHunters. 

The company also uncovered network campaigns originating in Russia, Iran, Turkey, and other regions that utilized AI to generate automated, authentic-looking social media profiles to push political narratives.

According to the findings, the vast majority of misuse occurred on older systems, including Claude Opus 4 and Claude Sonnet 4.5. Anthropic clarified that those earlier releases "were well below the threshold where they could meaningfully assist a sophisticated user in carrying out dangerous biological research."

"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," the report said. 

"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."

To address these vulnerabilities in newer releases like Claude Fable 5, the firm stated it has implemented "stronger safeguards that restrict access to a wide range of dual-use biological research queries".

The company added that sharing threat data publicly serves a broader protective purpose:

"We’re publishing this work because we believe we have a responsibility to disclose malicious misuse of our services. As models become increasingly capable, their risks will increase, unless AI developers and society’s defenders act to make them safer.”

"We hope that the findings in this report will help other developers recognise similar patterns on their own platforms, give governments and civil society a clearer view of how emerging threats take shape, and strengthen collective defences," Anthropic said.

The report's release coincides with heightened internal and external scrutiny regarding AI safety. Resigning researcher Jacob Coxon warned publicly that leading labs "are racing straight to self-improving superintelligence and gambling with our lives."

Similarly, OpenAI Chief Scientist Jakub Pachocki recently advocated for voluntary industry pauses, stating:

 "I am concerned no one is prepared for the consequences of a continued rapid rise in machine intelligence.”

In response to growing safety warnings, US Senator Bernie Sanders introduced legislative measures to temporarily halt advanced development, saying "When scientists tell you there is a chance, a chance that it could have a cataclysmic impact on humanity, you've got be a moron not to say, slow it down," on Thursday on BBC's Newsnight.

On the other hand, President Donald Trump rejected calls for a development pause, arguing that "if we don't win AI, we're going to be put in a very bad position"