90 Countries Are Building AI Governance. The US Isn’t One of Them.

The Center for AI and Digital Policy released the sixth edition of its AI and Democratic Values Index on April 9, covering 90 countries, up from 80 last year. The report runs 1,700 pages with more than 8,500 footnotes, drawing on nearly 1,500 contributors from over 120 countries.

Six countries reached the top tier: Canada, Japan, the Netherlands, Norway, Switzerland, and the United Kingdom. The United States was not among them.

That finding arrived the same week Stanford’s HAI released its own annual report, the 2026 AI Index. Stanford found that the US ranks 24th in the world in AI adoption, at 28.3%, far behind countries like the UAE and Singapore. The US leads the world in AI investment and model development. It ranks in the lower half of the countries surveyed in how widely its own population actually uses the technology.

Stanford’s report also found that AI experts and the American public disagree on nearly everything about AI’s future. A full 73% of AI experts said AI’s impact on jobs will be positive. Only 23% of the public agreed. And only 31% of Americans said they trust the government to regulate AI responsibly, the lowest of any country Stanford surveyed.

These are two different reports measuring different things. The CAIDP Index evaluates whether countries are building governance aligned with human rights, transparency, and democratic oversight. The Stanford report tracks adoption, investment, and public opinion. But they land in the same place. The US builds more AI than anyone and governs it less than most of the countries tracking alongside it.

The CAIDP report is direct about why. It describes the US government’s revocation of Biden-era AI safety protections as “the beginning of a steady decline in the country’s leadership in trustworthy artificial intelligence and global governance.” The report documents the US withdrawal from UNESCO, opposition to the UN AI Scientific Panel, and efforts to weaken EU AI Act protections through the proposed AI Digital Omnibus.

According to CAIDP, the administration also dropped civil society, think tank, and labor engagement groups from the US-hosted G20 summit in December, keeping only the business group. CAIDP President Merve Hickok warned in the organization’s newsletter that “industry is now deeply embedded into government policymaking, while civil society and academia have been sidelined.”

This is not happening in a vacuum. In July 2025, the White House signed an executive order banning federal agencies from buying AI models that incorporate diversity, equity, and inclusion principles. In December 2025, a follow-up order directed the FTC to build the case for overriding state AI laws, including anti-discrimination rules, under federal consumer protection law. In March, I wrote about how the White House’s national AI policy relies on oversight mechanisms that do not yet exist. The CAIDP Index suggests the gap has grown wider since then, not narrower.

Meanwhile, 45 countries now support the Council of Europe’s AI Treaty, the first binding international agreement on AI. New AI laws took effect this year in China, Japan, Peru, South Korea, and Vietnam. The CAIDP report found that the countries making the most progress share a few things in common. They are building independent oversight bodies. They are requiring transparency about how AI systems make decisions. And they are creating ways for people to challenge those decisions when they go wrong.

At the federal level, the US is moving in the opposite direction. The administration is pushing to override state AI safeguards and framing anti-discrimination rules as obstacles to innovation.

There is a version of this story where deregulation is a competitive advantage. The administration would say the US is leading by getting out of the way. But the Stanford data complicates that. Americans are not adopting AI at the rate the investment would predict. They do not trust their own government to manage it. And the people building the technology and the people living with it are further apart in how they see its impact than they have ever been.

The US built more AI systems than any country in the world last year. It also built less governance infrastructure to manage them than Canada, Japan, Norway, the Netherlands, Switzerland, or the UK. At some point, the distance between those two facts stops being a policy choice and starts being a warning.