Anthropic, the US-based AI company behind the Claude model, has published a report detailing how Russian-linked actors exploited its technology for cyber-espionage, disinformation, and the development of autonomous drone swarms. The findings, released this month, highlight a growing trend of state-aligned groups leveraging commercial AI tools for hostile operations.
According to the report, Anthropic blocked all identified malicious activity, used the findings to bolster its security systems, and shared intelligence with government bodies and industry partners. The company emphasized that its actions were part of a broader effort to counter AI abuse.
Espionage against Ukrainian and European targets
One of the most significant cases involved a hacking group designated GTG-20006, which used Claude to automate operations against more than 20 organizations. Targets included government ministries, defense and intelligence agencies, embassies, think tanks, and defense-industrial companies. Ukraine was the primary focus, with suppliers of military drone technology and their supply chains also attacked. The group additionally targeted government departments in the Middle East and Asia with maritime connections.
Anthropic linked GTG-20006 to Midnight Blizzard, a hacking group attributed by Western intelligence to Russia's foreign intelligence service (SVR). The group used Claude in multiple ways: as an engineering assistant for writing malware, phishing kits, and surveillance tools; as an executor of commands on victims' networks, including credential collection and data extraction; and in fully autonomous operations with minimal human oversight.
This incident underscores the evolving threat landscape in Europe, where cyber-espionage remains a persistent concern. As NATO jets down drones over Lithuania and tensions persist along the eastern border, the use of AI in such attacks adds a new layer of complexity for defense forces.
Disinformation in the Central African Republic and Moldova
Anthropic, working with the All Eyes On Wagner project, identified and removed an account operated by a Russian-speaking individual in Bangui. This person ran a disinformation operation for Radio Lengo Songo, a station established in 2017 with funding from the Kremlin-linked Wagner group. The operator prepared daily pro-Russian content in coordination with Russian state outlets RT, Sputnik Afrique, and TASS, as well as the local "Russian House" cultural center, part of a network used for political influence abroad.
The content consistently promoted the CAR government and Wagner while attacking France and domestic opposition figures. The operation was designed to appear locally run, with contracts and scripts presented as the work of a Bangui-based radio station. Investigators identified the individual through invoices for a Claude subscription issued in central Madrid, linking them to Africa Politology, also known as "The Company," the information and political technology arm of the African Corps, the successor to Wagner in Africa, reportedly controlled by the SVR.
In Moldova, a former regional head of the Russian state-run Sputnik network used Claude as a content-production tool, adapting news, polling data, and opposition material to fit pro-Kremlin narratives. This activity aligns with broader Russian disinformation efforts in the region, which have been denied by the EU in other contexts.
Autonomous drone swarm development
Anthropic also identified and shut down a group of Russia-linked developers who used its Claude Code tool to write software for kamikaze drones, including an autonomous targeting system tested on real hardware. The developers trained AI models on footage from combat operations in Ukraine and used a fixed location in the Donetsk region as a demonstration strike target. The project's stated goal was a fully autonomous swarm of FPV kamikaze drones.
Anthropic linked nine accounts involved in the project to a regional Russian university and a federal research center connected to the Russian Academy of Sciences, though it did not name either institution. This development is particularly concerning given the ongoing drone strikes on civilian targets in Ukraine, which have killed dozens of people.
Claude is officially unavailable in Russia, with access to Anthropic's website and API blocked and Russian bank cards not accepted for payment. The developers bypassed these restrictions by routing traffic through commercial virtual private servers (VPNs).
The report highlights the dual-use nature of AI technology and the challenges of preventing its misuse. As European nations grapple with Belarus re-emerging as a launchpad for Russian drone strikes, the need for robust AI governance and international cooperation has never been more urgent.
