Generative AI is changing how state-linked influence operations are run, according to a Lawfare analysis of threat reports from frontier AI companies and platforms. Iranian propagandists have used Claude to draft doctrine and plan events, Russian state media staff have used it to produce articles and on-air chyrons, and China-linked operators have deployed ChatGPT to generate political cartoons and false claims about AI companies. The common thread is production: AI lets actors write, translate, and imitate target audiences with fewer telltale mistakes than earlier troll-farm efforts.
But the reports also describe a deeper shift in operations. AI is moving beyond content creation into the back office—handling audience research, coding login automation, scoring reporters against rubrics, and even managing radio station staff. The Lawfare piece argues that the IRA's old stand-up meetings have effectively become prompts, and that agentic AI is scaling operations the way generative AI scales content. One operator can now do more jobs, and coordination can run through software rather than human handoffs.
The harder question, the source notes, is persuasion. AI makes influence apparatuses cheaper, faster, and more persistent, but it cannot manufacture trusted distributors. There is abundant evidence of increased capacity, but much less evidence that AI-generated campaigns are actually changing public opinion. The article stops short of concluding these operations are ineffective, but it stresses that capability gains and persuasive impact are not the same thing. As this is a single source, no contrasting views are available{