Governments are turning to Claude to automate spying

# Analysis The emergence of state actors weaponizing Claude for surveillance represents a critical inflection point in how authoritarian regimes conduct intelligence operations—not because AI created these capabilities, but because it dramatically reduces the technical barriers and operational costs of mass monitoring. Mali, China, and Iran can now automate the analysis of communications, social media, and movement data at scales that previously required larger teams of human analysts, meaning governments can surveil dissidents and political opponents with fewer resources and less institutional friction. The fact that Claude bans haven't deterred these operations underscores a harder truth: the technological genie cannot be contained through single-vendor restrictions when multiple frontier AI models offer similar capabilities, and regimes with sufficient resources will simply migrate to alternative platforms. For civil society organizations, journalists, and activists in these countries, this means threat models must fundamentally shift to assume that their digital footprints are subject to automated, continuous analysis by state actors—requiring new approaches to operational security that many independent media outlets and NGOs are unprepared to implement.