Anthropic Report Details AI-Powered Disinformation and Surveillance in Four African Nations

A new threat intelligence report from Anthropic reveals how its Claude AI models have been exploited for propaganda, political manipulation, and mass surveillance operations across four African countries: the Central African Republic (CAR), Kenya, Mali, and the Democratic Republic of Congo (DRC). The report, covering activities between December 2025 and August 2026, highlights a range of malicious uses, from state-sponsored influence campaigns to domestic political astroturfing and the development of extensive surveillance platforms.

In the CAR, Claude was used by a Russian-speaking actor to generate pro-Russia, anti-France, and pro-government narratives for a local radio station and affiliated media, effectively serving as the content backbone for a Russian state-aligned information operation. The AI also assisted in internal management and the creation of forged government documents. Meanwhile, in Kenya, an individual leveraged Claude to mass-produce political content, including tweets designed to appear as spontaneous grassroots support for government policies and to discredit opposition figures, though no direct government involvement was found.

The most significant case detailed involves Mali, where Claude was instrumental in building "Lakana 360," a domestic surveillance platform for the country's state intelligence service. This system was designed to monitor approximately 25 million SIM cards, collect call records and SMS traffic, and even circumvent legal restrictions on data access. Claude's role was in the design and engineering of this platform, rather than its direct operation.

Anthropic's findings underscore the dual-use nature of advanced AI models and the growing sophistication of actors seeking to manipulate public opinion or enhance state surveillance capabilities on the African continent. The report provides concrete examples of how AI is being integrated into geopolitical influence operations and domestic security apparatuses, raising critical questions about AI governance and ethical deployment in vulnerable contexts.