The company also accused several China-based AI laboratories of attempting to extract Claude's capabilities and said that some Chinese firms routed live customer conversations through their systems for training purposes

SAN FRANCISCO, California: Anthropic said it disrupted attempts to use its Claude AI models for biological weapons research, weapons development, and cyberattacks, including a suspected Russia-linked espionage campaign targeting Ukrainian officials.

The findings were detailed in Anthropic's latest Threat Intelligence report, covering malicious use of its technology over the past eight months.

Anthropic documented five cases in which scientists used Claude in ways that could support biological weapons development. In one case, a researcher in a region not supported by Anthropic used virtual private server infrastructure to access Claude and spent weeks planning avian influenza mammalian-adaptation experiments.

Anthropic said unsupported regions include countries such as Russia, China and North Korea. It banned accounts involved in the activity and incorporated the findings into its safeguards, enforcement and threat-intelligence systems.

The company did not identify the institutions or countries involved, or specify the biological agents and research techniques used.

Anthropic also said Claude had been misused to "develop software for conventional weapons, including firearms, missiles, armed drones, bombs, and other munitions, as well as the targeting and control systems that operate them."

The report cited operators in China, Russia, and Yemen who used Claude for weapons design and development or for intelligence gathering and procurement linked to weapons programs.

Rapid advances in Anthropic's models have increased the risks, according to Jacob Klein, the company's head of threat intelligence.

"A year ago, let's say you wanted to optimize a drone or optimize the software on a missile, the models just wouldn't be as good at that task as they are now," he said.

Anthropic also said cybercriminals and state-backed hackers were increasingly using AI to automate large portions of attacks.

"The use of AI went beyond simple questions and responses from a chatbot but rather involved the use of multi-agent frameworks," it said.

One hacking group allegedly used AI in phishing campaigns, hotel Wi-Fi hijacking and WhatsApp takeovers targeting Ukrainian government, military and diplomatic officials. Anthropic said the group's methods were consistent with those of Midnight Blizzard, a Russia-based threat actor previously linked by the U.S. government to Russia's SVR foreign intelligence service.

The Russian Embassy in Washington did not immediately comment.

Anthropic said the group used AI to create a system that could detect when security software identified malware and repeatedly rewrite the code until it evaded detection.

The company also said it disrupted activity linked to affiliates of the ShinyHunters cybercrime collective.

Separately, Anthropic said it stopped attacks from seven China-based AI labs, including Alibaba, Moonshot, DeepSeek and Xiaomi. The companies did not immediately respond to requests for comment.

Anthropic said operators linked to Alibaba carried out what it described as the largest "illicit distillation" attack, allegedly seeking to extract Claude's capabilities to improve Alibaba's Qwen models.

The company said it observed more than 151 million exchanges attributed to Alibaba between May and July 2026, peaking at nearly 3 million a day from more than 3,500 accounts it described as fraudulent.

Anthropic also said Moonshot and DeepSeek allegedly routed live customer conversations, sometimes containing sensitive information, through Claude and used its responses as training data.

China's foreign ministry said it was unaware of the report and maintained that AI should be developed for good, while opposing what it called distortion of facts and smears against the country.