Anthropic report: Is Russia using AI for disinformation in the Central African Republic and elsewhere?

A recent report by Anthropic suggests that Russia may be utilizing the AI assistant Claude to disseminate disinformation in the Central African Republic and other nations. The report, titled "Detecting and Countering Misuse of AI," was published in September by the US-based company. According to the report, Claude was employed to create propaganda content and support disinformation campaigns, primarily for Radio Lengo Songo, a radio station reportedly funded by Russia.

The content generated by Claude portrayed the Central African government and Africa Corps mercenaries positively, while criticizing France and the opposition. Lou Osborn, a member of All Eyes on Wagner, an investigative collective, played a role in the research for the report. He believes these tactics are part of a long-standing practice associated with the Wagner Group in the Central African Republic.

Since 2016, Faustin-Archange Touadera has been the president of the Central African Republic. The report also revealed that Claude was prompted to depict opposition figures as armed militants, aiming to provoke security forces' response. Although Claude initially refused this request, the report did not disclose the names of the opposition figures monitored or the specific data collected.

This revelation highlights the Central African Republic government and its Russian allies' determination to eliminate opposition figures. Martin Ziguele, chairman of the Movement for the Liberation of the Central African People, acknowledged similar activities dating back to 2020 when the CPC, a coalition of rebel groups, attacked Bangui.

These attacks were often falsely linked to opposition leaders and alleged foreign funding. A proposed law targeting foreign agents in the country was also part of these efforts. The report further alleges that Russia denies these claims, labeling them as vague and unfounded. However, it indicates that the Central African Republic is not an isolated case, with similar activities reported in the Democratic Republic of Congo, Kenya, Sudan, and Mali.

In the DRC, Anthropic discovered 318 articles published by a network of around 70 fake news sites, distributed through over 250 fake accounts on X, supporting the Congolese government's position against Rwanda. In Kenya, Claude generated 50 hostile messages against the opposition at a time, while in Sudan, AI was used to create fabricated witness statements for the United Nations.

In Mali, an intelligence service adviser allegedly used Claude to develop a platform for monitoring telecommunications, designed to target 25 million SIM cards without judicial authorization. Anthropic uncovered and dismantled these operations between December 2025 and August 2026 by deleting accounts and strengthening its detection systems.