Anthropic, the artificial intelligence company behind the Claude chatbot, announced findings that a United Arab Emirates-linked organization operated an extensive disinformation campaign targeting United Nations experts working on Sudan. The group used Claude to generate a wide range of materials — including research reports, advocacy documents, social media posts, and written testimony — all designed to shape expert opinions and policy recommendations regarding the ongoing conflict in Sudan.
According to Anthropic's investigation, the operation was primarily aimed at discrediting the Muslim Brotherhood and its Sudanese affiliate, the National Unionist Party, by framing them as the primary cause of instability in the country. The generated content consistently pushed a narrative aligned with the UAE's strategic interests, which have been widely reported to support General Mohamed Hamdan Dagalo, known as Hemedti, and his Rapid Support Forces during Sudan's civil war.
The scope of the AI-generated output was extensive. Documents prepared through Claude were crafted to appear as independent analysis and credible testimony before UN bodies. Social media content was produced to amplify these narratives across digital platforms, creating an impression of grassroots support for the desired position. Advocacy materials followed a similar pattern, presenting AI-drafted arguments as the work of local activists and Sudanese civil society organizations.
Anthropic stated that it detected the unusual volume and origin of the requests during its normal operations and subsequently launched an internal review. The company emphasized that while the content was generated using its technology, the underlying strategy, direction, and oversight came entirely from the external group operating the accounts.
Sudan has been embroiled in a devastating civil war since April 2023, pitting the Sudanese Armed Forces against the Rapid Support Forces. The conflict has killed tens of thousands and displaced millions. International actors, including the UAE, have faced accusations from human rights groups and UN officials of intervening in ways that prolong the fighting rather than facilitate peace.
This revelation adds to growing concerns about how foreign governments and affiliated entities are exploiting readily available AI tools to wage information wars at the international diplomatic level. Legal and policy experts note that such campaigns exploit the speed and accessibility of generative AI to flood institutions like the United Nations with professionally written but entirely fabricated material.
Anthropic has indicated it is reviewing its usage policies and detection capabilities to prevent similar misuse in the future.
