Anthropic flags UAE-linked misuse of AI targeting the Muslim Brotherhood

Anthropic's report puts a specific name and tool to a type of influence campaign that governments and platform operators have worried about for years. The operation used Claude AI to run around 300 fake social-media accounts, set up a front NGO, ghost-write UN testimony, and build files on European lawmakers, journalists, and UN officials. Anthropic linked the activity with high confidence to UAE government officials, an attribution that carries weight because it comes from the AI developer whose system was used. The UAE has designated the Muslim Brotherhood a terrorist organization since 2014 and made countering political Islam a pillar of its regional policy. The United States treats the Brotherhood's Egyptian, Jordanian, Lebanese, and Sudanese branches as terrorist organizations. The report also noted that this campaign was not the most troubling example of AI misuse it found. Anthropic separately uncovered Russian cyber attacks, Chinese surveillance of Uyghurs, and Yemeni-developed software for armed drones, bombs, missiles, and other weapons. For AI developers, the stakes appear to hinge on how quickly they can detect and report such state-linked activity, and on whether their disclosures lead to clearer expectations from regulators and enterprise customers about misuse monitoring.