A covert influence operation that Anthropic says was directed by United Arab Emirates officials allegedly used the company’s Claude AI chatbot to run a global campaign against the Muslim Brotherhood.
According to a threat intelligence report published by Anthropic on September 10, the operation used AI to create a fake human rights organisation, prepare testimony for the United Nations, build fake social media accounts and compile profiles on European lawmakers and journalists.
The operation was one of nine cases examined in the report, which looked at the misuse of Anthropic’s AI models between December 2025 and August 2026.
Anthropic Says UAE Officials Directed Operation
Anthropic said it assessed with “high confidence” that the campaign was directed by UAE government officials. The company said an internal document linked to the operation identified senior UAE figures as the intended recipients of the work.
The UAE has not publicly responded to the allegations. The claims have also not been independently verified.
Anthropic said the findings represent its own investigation because the company’s AI product was used in the operation. It said the account involved was banned and that additional safeguards were introduced. The company also said it shared information about the operation with other organisations.
AI Used to Build Covert Campaign
According to Anthropic, one operator used Claude to maintain an AI persona called “Deadshot”. The persona was created around instructions to carry out what was described as a coordinated campaign to dismantle the Muslim Brotherhood globally.
The operation allegedly involved around 300 fake social media influencer accounts. It also included a front organisation that copied the identity of a real Swiss human rights group and published reports presented as independent material.
The operator also used Claude to prepare testimony intended for the UN Human Rights Council. Anthropic said the testimony was written in a way that did not mention the UAE.
European Lawmakers and Journalists Profiled
The campaign also involved detailed profiles of 18 members of the European Parliament and several journalists, according to Anthropic.
The operation allegedly created what Anthropic called “counter accountability” files on UN special rapporteurs who had criticised the UAE’s conduct in Sudan’s war.
The company said Claude was used to turn political doctrine and other internal files into intelligence style briefs, reports using copied identities, UN testimony and profiles of individuals.
Some of the material was reportedly prepared for delivery to senior UAE officials.
Fake Independence Used to Hide State Links
Anthropic said internal documents described the apparent independence of the social media network as its “greatest strategic asset”.
The company said this suggested that the operation was designed to hide any connection to state actors.
In another case, the operator allegedly used the identity of a real Sudanese rights group to submit testimony to the UN. The testimony was presented as coming from an independent local witness, despite the group’s connection to the conflict in Sudan.
Anthropic classified the operation as Category Three, a mid level assessment of influence campaign reach.
The company said it could not confirm whether the testimony or profiles reached their intended audiences or influenced policy decisions.
