A single operator used an AI persona called "Deadshot" to generate intel briefs, fake UN testimonies, targeting lists of EU parliament members and journalists, and content for ~300 fake influencer accounts — all designed to be deniable and delivered to senior UAE officials.
They cloned a real Swiss/Sudanese human-rights NGO's identity, drafted speeches for the UN Human Rights Council with instructions to never mention the UAE, and built counter-dossiers on UN officials who'd criticized UAE actions in Sudan.
Anthropic banned the accounts, rated it Category Three (multi-platform, no confirmed mass public impact), and published the findings in its September 2026 threat report.
The UAE has a long-running campaign against the Brotherhood dating to 2014, and notably has also invested in Anthropic through MGX.
Anthropic caught a UAE-linked operation using Claude to run an influence campaign against the Muslim Brotherhood and Sudan.
29
17
8
4
3
2
2
1September 11, 2026 4.7K 43
