> ## Content Index > Fetch the complete content index at: https://aiadoption.org/llms.txt > Use this file to discover other available public pages before exploring further. # 79% globally want AI disclosure — but only 31% of Americans trust their own government to regulate it - URL: https://aiadoption.org/ai-analysis/79-globally-want-ai-disclosure-but-only-31-of-americans-trust-their-own-government-to-regulate-it/ - Published: 2026-09-21T05:42:03.000Z - Updated: 2026-09-21T05:42:03.000Z - Description: 79% of respondents across 30 countries want companies to disclose their AI use, but only 31% of Americans trust their own government to regulate AI responsibly — the lowest of any country surveyed. The disclosure expectation is global; the enforcement trust isn't. - Author: Jassie - Tags: AI Analysis, AI Compliance, AI Governance, AI Policy 79% of people across 30 countries say companies using AI should be required to disclose it (Ipsos AI Monitor 2025, via Stanford AI Index 2026, 9.1). Only 31% of Americans trust their own government to regulate AI responsibly — the lowest of any country surveyed (9.3). That is the compliance environment in one contrast: near-consensus public expectation of disclosure, and — in the US — the weakest public trust in the institutions positioned to enforce it. Ipsos AI attitudes tracker, 2022–25 — 79% say AI use should be disclosed (2025) ## What the data shows The headline read of the 79% is that disclosure regulation is inevitable, so waiting for the mandate is a defensible posture. The reality the same survey shows is different: the expectation is global, but the enforcement trust isn't. An organisation that waits for a trusted regulator to formalise the norm will, in the US at least, be waiting on an institution that two-thirds of the public doubts. The disclosure expectation itself is striking for its consistency. Across all 30 countries Ipsos surveyed (n = 23,216), a majority say AI use should be disclosed, and the 79% global average is up from 67% in 2022 (Ipsos AI Monitor, 2022–2025). The trajectory is rising, and the consensus is broader than for almost any other AI-related question in the survey — it crosses sentiment, demographic, and political lines. One caveat keeps the reading honest. A 79% stated expectation is not the same as 79% willingness to punish non-disclosure — survey expectation and market behaviour diverge, and Ipsos measures the former. What the number does establish is the direction and breadth of the norm: no surveyed country recorded a minority expectation, and the figure has moved 12 points in three years. Treating it as a floor, rather than a forecast of enforcement, is the defensible read. Trust in government to regulate AI is the fragmented dimension. The global average is 54%. Southeast Asia leads: Singapore 81%, Indonesia 76%, Malaysia 73%, Thailand 70%. China and Mexico sit at 67%, India at 65%. The trailing group: the United States 31%, Japan 32%, Hungary 33%, Great Britain 39%, Canada 40%, France 42% (Ipsos AI Monitor 2025, via Stanford AI Index 2026, §9.3). Trust in own government to regulate AI by country, 2025 ## Three drivers explain the US outlier. **The first is political polarisation of US AI policy.** The January 2025 rescission of Executive Order 14110 and the deregulatory pivot that followed happened in public view. Citizens who backed the precautionary framework now distrust the new one; citizens who backed deregulation distrusted the old one — divided trust pulls the average down. **The second driver is that institutional trust generally runs lower in the US than in most surveyed countries.** The 31% figure tracks downstream from broader American scepticism toward federal institutional capability; AI regulation is one specific instance of a wider pattern, not an AI-specific judgement. **The third is proximity.** When the most prominent AI developers operate from the same jurisdiction as the regulator, respondents have more reason to doubt the regulator's independence than in countries whose dominant AI providers operate from abroad. The closeness of "the AI companies" and "the regulator" is itself a trust cost. ## Why this matters for mid-market For a mid-market firm, the 79% expectation means disclosure is now the floor of acceptable practice, not a differentiator. Disclosing AI use prominently — in customer communications, employee communications, and product interfaces — matches the expectation. Disclosing only where legally required operates against it, and bears the trust cost with the four in five customers who expect it regardless of what the law compels. In the US specifically, organisational transparency substitutes for institutional trust. With 31% trusting federal regulation, the public looks to the organisation itself for AI transparency rather than expecting it via government enforcement. Matching only the legal minimum ties your credibility to a regulatory system the public doesn't trust; meeting or exceeding what regulation would require makes the organisation its own credibility source. This is one of the few compliance postures where a mid-market firm can genuinely move faster than a Fortune-500 competitor: one disclosure standard, applied everywhere, is achievable without a dedicated policy team, while the larger firm reconciles the positions of forty product lines. The operational catch is jurisdictional fragmentation. EU AI Act disclosure requirements are converging toward specific content and format standards, while US state requirements are fragmented across California, Texas, Colorado, Utah and others. A generic "we use AI" notification satisfies the public expectation but not necessarily the specific regulatory requirements. Disclosure infrastructure therefore needs a jurisdictional adaptation layer — at mid-market scale that can be a maintained register of where you operate, what each jurisdiction requires, and which disclosure text applies, owned by one named person. It does not need to be a platform. > The structural observation behind all of this: the gap between global disclosure expectation (79%) and US institutional trust (31%) is not closing. The expectation keeps rising; the trust environment depends on broader political dynamics that are not AI-driven and show no sign of resolving. Disclosure strategies that engage with the mismatch now will be better positioned than strategies that wait for the trust environment to improve. ## What changes Monday morning Ask whoever owns customer communications for the current inventory of customer-facing AI touchpoints, and check each against one standard: is the AI use disclosed where the customer encounters it? Where it isn't, add the disclosure this quarter rather than waiting for a mandate — and log the decision, the jurisdictions it covers, and the owner of the register in the same document. The expectation is at 79% and rising, the US enforcement environment is distrusted and fragmented, and the mismatch will persist through 2026–2028\. The cost of proactive disclosure is modest against the cost of being the non-disclosed brand once the norm is fully established. --- ### Sources - **Primary**: Stanford AI Index 2026, Chapter 9 (Public Opinion) 9.1 and 9.3 — [hai.stanford.edu/ai-index/2026](http://hai.stanford.edu/ai-index/2026?ref=aiadoption.org) - **Disclosure and trust data**: [Ipsos AI Monitor](https://www.ipsos.com/?ref=aiadoption.org), 2025 — 30-country survey of 23,216 adults; disclosure expectation and government-trust-to-regulate-AI measures - **Trajectory data**: Ipsos AI Monitor, 2022–2025 — 79% disclosure expectation up from 67% in 2022 - **Regulatory context referenced**: US federal AI policy reversal (January 2025 rescission of EO 14110); EU AI Act; US state-level disclosure requirements (California, Texas, Colorado, Utah)