Russia Uses Claude AI for Espionage and Disinformation

Anthropic has released a new report detailing how its Claude AI model was exploited by a Russian‑linked hacking group to automate cyberattacks against more than 20 organisations. The report lists targets that include Ukrainian defence ministries and drone suppliers, as well as other entities in the region. The attackers used the model to streamline phishing, credential‑stealing, and other intrusion techniques, allowing them to scale their operations rapidly.

In addition to the direct attacks, the report documents separate operators employing Claude to generate pro‑Kremlin disinformation campaigns in Africa and Moldova. These campaigns involved the creation of synthetic media and coordinated messaging designed to influence public opinion and political discourse in those countries. Anthropic’s findings underscore the growing risk that advanced generative AI can be weaponised for both cyber intrusion and information manipulation.