A new report released by the artificial intelligence firm Anthropic has outlined specific ways in which Russian-affiliated entities are leveraging AI technology to bolster their intelligence-gathering and cyber-warfare capabilities. According to the findings, these actors are increasingly using large language models to automate tasks related to espionage, such as drafting deceptive communications and analyzing large datasets to track political dissidents.

Anthropic’s analysis suggests that the accessibility of generative AI has lowered the barrier to entry for sophisticated cyber operations. By automating the creation of convincing phishing emails and social engineering scripts, state-linked groups can scale their efforts to monitor and target individuals perceived as threats to the Russian government. The report emphasizes that while these tools are not inherently malicious, their misuse by state actors presents a significant challenge to global cybersecurity.

While the report provides a technical breakdown of these activities, it stops short of detailing specific successful breaches or the full extent of the damage caused by these AI-assisted campaigns. The findings serve as a warning to the tech industry and government agencies regarding the dual-use nature of AI, urging for more robust safety protocols to prevent the exploitation of these models for state-sponsored surveillance and digital attacks.