Anthropic reveals Russia-linked groups misused Claude chatbot for war efforts and propaganda

AI developer Anthropic has disclosed in its latest report several cases in which Russia-linked groups misused its Claude chatbot for harmful purposes.

Cyberattacks on Ukrainian institutions

One of the most serious cases involved a hacking group that used Claude to automate cyberattacks against more than 20 organizations, primarily in Ukraine. The attacks targeted state, defense and intelligence institutions, as well as suppliers of military drone technology. The chatbot was used to develop malware and phishing tools and to extract information from compromised systems.

Propaganda production across several countries

Claude was also used to produce content favorable to Russia. In the Central African Republic, it was used to help generate content for a radio station linked to the formerly Kremlin-backed Wagner group. In Moldova, a former regional head of Russian state media outlet Sputnik used the chatbot to adapt news and poll data into pro-Kremlin messaging.

Development of autonomous drones

An especially concerning case involved a Russia-linked group that used the Claude Code tool to develop software for kamikaze drones, including an autonomous target-identification system. The AI models were trained using video footage gathered during the war in Ukraine, with the project's stated aim being to build a fully autonomous swarm of FPV kamikaze drones.

Access by bypassing restrictions

Claude is not officially available in Russia, but the users involved bypassed the restrictions using virtual private networks and other servers. Anthropic said it has blocked all the misuse cases identified in the report, used the findings to strengthen its own security systems, and shared the information with government agencies and industry partners.