A new report from the AI safety company Anthropic has identified instances where its Claude AI model was leveraged by Russian-linked hacking groups to support malicious activities. According to the findings, these actors used the technology to automate cyberattacks against more than 20 organizations. The targets of these digital intrusions included various Ukrainian defense ministries and companies involved in the supply chain for drone technology.

Beyond cyber espionage, the report highlights that separate operators utilized the AI to generate and disseminate pro-Kremlin disinformation campaigns. These influence operations were primarily focused on regions including Africa and Moldova. Anthropic’s investigation suggests that the AI was used to streamline the creation of content and technical tasks that support these broader geopolitical objectives. While the report outlines the specific nature of the abuse, it serves as a broader indicator of the challenges AI developers face in preventing their tools from being repurposed for state-sponsored digital warfare and propaganda efforts. The company has indicated it is taking steps to mitigate such misuse, though the incident underscores the ongoing tension between AI accessibility and security risks.