Survey finds that 76% of Chinese citizens and 68% of Indians say they trust AI companies compared to only 35% in France, the UK, and the US
Madhumita Murgia / Financial Times : Tweets: @reema__patel , @ethicsinai , and @ftweekend Tweets: Reema Patel / @reema__patel : I comment on the implications of a new @IpsosMORI survey on #AI for the @FT @madhumita29. These findings reveal the gap between people, policymakers and developers and suggest the need for a Warnock style deliberative inquiry. https://www.ft.com/... @ethicsinai : “Globally, about half said AI technology has more benefits than drawbacks, while two-thirds felt gloomy about its impact on their individual freedom and legal rights.” https://www.ft.com/... @ftweekend : 'Research underlines the importance of AI designers recognising that we don't all belong to one homogenous population, with the same understanding of the world.' Tap here for @madhumita29's latest column: https://www.ft.com/...
Context & Ripple Effects
The Ipsos MORI finding lands mid-arc in a trust divide that keeps reappearing across surveys: four years earlier, a CCTV–Tencent poll had 76% of Chinese respondents calling AI a privacy threat, so Chinese trust in AI companies coexists with sharp privacy concerns rather than replacing them. The same pattern resurfaces later, when a UN survey of 21 countries confirms trust in AI is strongest in low-income and developing nations — 83% in China versus 37.5% in the US.
Reema Patel's read on the data is that the gap runs three ways — between people, policymakers, and developers — and she argues for a Warnock-style deliberative inquiry to close it. The commercial stakes are real: a market where two-thirds of citizens trust AI companies is one where deployment faces less social friction than the 35%-trust markets of France, the UK, and the US.
First-order effects
- AI companies operating in China and India get a legitimacy head start — with 76% and 68% trust respectively, consumer-facing deployments there meet less resistance than in France, the UK, and the US, where only 35% extend the same trust.
- Western policymakers are handed evidence that their publics distrust the very firms building the technology, sharpening Patel's argument for a deliberative process like the Warnock inquiry rather than developer-led self-governance.
Second-order effects
- Low-trust Western markets push governments toward sovereign alternatives — France's response path includes €655M in AI investment through 2030, a planned Mistral-powered state chatbot, and roughly 10 GW of new computing capacity, effectively substituting state-backed providers for distrusted private ones.
- The trust asymmetry shapes product strategy: firms can iterate faster in high-trust Asian markets and treat the US, UK, and France as regulatory-first markets, a split visible in later polling where 52% of Americans and 63% of Britons say they would be uneasy with mostly AI-produced news.
Third-order effects
- If the pattern holds, AI adoption diverges structurally by region: high-trust markets normalize everyday AI use first while low-trust democracies build deliberative and regulatory gatekeeping, entrenching different governance regimes around the same technology.
- The persistent people-policymaker-developer gap points toward institutionalized citizen assemblies on AI becoming standard democratic infrastructure, the way the Warnock model was invoked for assisted dying.
The trend: Global trust in AI is settling along development lines — highest where AI arrives as a practical tool, lowest in wealthy democracies — and that divide is increasingly deciding which markets deploy first and which regulate first.