Iranian state propagandists used Anthropic’s Claude to draft doctrine for their influence campaigns and plan the Supreme Leader’s funeral, exposing how foreign adversaries now employ artificial intelligence for administrative and operational automation rather than just producing simple text.
Iranian and Russian State Media Adopt AI for Propaganda
State-linked actors across Iran, Russia, and China have moved far beyond basic content generation. According to threat reports from frontier AI companies, Iranian propagandists utilized Claude to draft the doctrine for their influence campaigns. In Moldova, a former Sputnik Moldova editor-in-chief used AI to churn out articles and push fabricated claims about the country’s president ahead of the 2025 election. Meanwhile, a staffer at a Russian state media outlet relied on AI to write on-air chyrons.
These productivity gains allow troll farms to scale output while avoiding typical errors. Automated systems handle background research, timing strategies, and data processing without human fatigue. In the Central African Republic, a coordinator used Claude to draft contracts pledging loyalty to “Russia and its contingent,” score reporters’ articles against a rubric, and recommend whom to fire.
Chinese Operators Automate US Political Interference
Chinese tech contractors have deployed artificial intelligence to meddle in domestic American debates over AI infrastructure. OpenAI reported that accounts tied to Chinese operators used ChatGPT to generate political cartoons blaming data centers for rising electricity bills. Another cluster had ChatGPT write code to automate logins across platforms, while false claims circulated that ChatGPT had leaked user data.
Despite the operational efficiency, these specific influence campaigns failed to capture broad audiences. Both OpenAI campaigns earned “Category One” ratings on the breakout scale, meaning the generated content never spread beyond the operator accounts. Real distribution remains a critical hurdle for foreign actors who lack trusted intermediaries and established viewer bases.
Did you know? While AI tools can drastically reduce the overhead of producing thousands of articles across multiple languages, security researchers note that automated networks of fake news sites often draw almost no real human engagement.
Frontier Companies Track Automated Cyber Operations
Beyond propaganda, state-backed groups use AI models to enhance operations. Anthropic documented a Russia-linked actor taking over the WhatsApp accounts of at least two former senior Ukrainian officials, quietly exporting their conversations—rich fodder for a hack-and-leak operation.
In addition to technical intrusion, the same Russian-linked actor successfully took over the WhatsApp accounts of at least two former senior Ukrainian officials, quietly exporting private conversations to supply material for future hack-and-leak operations.
How Foreign Operators Use Influencers and Fake Personas to Build Trust
How do foreign operators solve the distribution problem when using AI?
When automated networks fail to build organic trust, operators try to borrow it. Some teams hire unwitting influencers with established fanbases, while others pose as everyday immigrants to lobby real YouTubers or ghost-write testimony for international forums like the UN Human Rights Council.
What specific tasks can agentic AI perform compared to generative AI?
Generative AI primarily scales content production by writing text and creating images. Agentic AI takes over complex workflows, such as writing automation code and managing personnel contracts, and executing multi-step tasks without continuous human direction.
Did the Chinese data center influence campaign impact US policy?
Independent researchers who analyzed the campaign found the campaign gained essentially no authentic reach or public engagement.
Related reading