China’s DeepSeek faces global bans and investigations over data security

DeepSeek’s rapid spread and equally rapid backlash show what happens when a powerful model grows globally before trust frameworks are in place. The company’s own privacy policy acknowledges that user prompts and uploaded files are stored on servers in China, which has triggered bans on government use in Australia, parts of Europe and Taiwan, store delistings in some markets, and data‑protection investigations. In the US, policymakers are openly considering export controls and access restrictions.

For the race to AGI, this is a case study in geopolitical fault lines hardening around AI. Open‑source and low‑cost models coming out of China are attractive to developers and users, but they collide with national‑security concerns in democracies worried about data exposure and influence operations. As models get more capable, these trust gaps will only widen unless there are verifiable assurances about data handling and governance—something that’s hard to build across adversarial jurisdictions.

The practical effect is growing fragmentation of the AI ecosystem: certain models will be de facto off‑limits in large markets, shaping training data, benchmarks and talent flows. AGI systems may thus evolve under quite different regulatory and cultural regimes, which will matter when the world debates how, and under whose rules, they should be deployed.