Anthropic Disrupts State-Backed AI Surveillance Operations Linked to China, Iran and West Africa

Artificial intelligence company Anthropic said Thursday it had disrupted multiple state-sponsored surveillance operations that misused its AI models, warning that governments and state-aligned groups are increasingly turning to AI to monitor ethnic minorities, dissidents and political opponents.

The company said it identified and shut down the operations between January and July. The campaigns originated in China, Iran and West Africa, according to Anthropic’s latest report on the misuse of its AI models.

The surveillance efforts targeted diaspora and dissident communities that have historically faced scrutiny from the governments involved. These included pro-democracy figures in Hong Kong, Tibetan and Falun Gong communities across Asia, as well as Iranian minority groups and opponents of the Iranian regime living abroad.

In one case, Iranian actors used AI to develop a method for identifying individuals through their social media accounts. In another, a contractor working for Malian national security authorities used Anthropic’s Claude model to design software supporting intelligence-gathering activities.

“AI is now being used in place of an engineering workforce,” Anthropic said in its report.

The company also said it had disrupted attempts to use its models to design weapons, conduct potentially dangerous biological research and develop dating scams.

Anthropic Accuses Chinese AI Firms of Data Distillation

The San Francisco-based AI company also accused Chinese AI developers of deceptively using Claude to generate answers for their own users while simultaneously extracting the resulting data to improve their AI models.

This process, known as distillation, involves using the outputs of one AI model to train another.

Anthropic said Chinese developers have previously been accused of using the technique without permission, effectively benefiting from resources developed by US AI companies such as Anthropic and OpenAI.

The company now alleges that Chinese firms Moonshot AI and DeepSeek secretly used Claude to generate responses for their customers.

“In one instance, over a ten-day period, Moonshot relayed almost 300,000 customer requests to Anthropic,” the report said. Anthropic alleged that the requests were routed through a network of 5,380 fraudulent accounts, most of which appeared to be based in Singapore and Japan.

Some of the requests reportedly contained sensitive user information, potentially raising concerns about privacy agreements.

“We do not know if Moonshot notified their customers that their requests were being rerouted to Anthropic and exposed to a third party,” Anthropic said.

The company said DeepSeek used similar techniques.

AI Misuse Linked to Biological Research

Anthropic also reported detecting and blocking activity involving research related to potentially dangerous biological agents. However, the company said the intentions of the researchers were unclear and did not identify them.

“The individuals implicated in these case studies are working scientists,” Anthropic said, adding that it did not claim they intended to cause harm.

The research involved the chikungunya virus, a highly pathogenic strain of avian influenza, virus families that include smallpox and mpox, venoms and other toxins.

Anthropic said identifying the researchers or their laboratories could potentially expose them to harm, and therefore withheld their identities.