What happened
Stanford HAI’s 2026 AI Index chapter on policy and governance reports wider adoption of national AI strategies, growing attention to AI sovereignty, sharp regional differences in data-localization measures and supercomputing infrastructure, and a large gap between public and private AI investment. The supplied source does not mention Becerra or substantiate the candidate headline’s claim that there are “hardly any” AI regulations.
The primary source is Stanford HAI’s 2026 AI Index chapter titled “Policy and Governance.” It says national AI strategies are expanding fastest among countries that had no formal AI policy five years earlier. More than half of newly adopted strategies in 2024 came from emerging economies, and the page says additional countries in sub-Saharan Africa, Central Asia and the Middle East had strategies in active development as of 2025.
These are claims and measurements presented by Stanford HAI; the supplied excerpt does not provide the underlying methodology or country-by-country list. The chapter defines AI sovereignty as the goal of gaining more agency over domestic AI capabilities and reports that it is becoming a central principle of national AI policy. It says Europe and Central Asia increased state-backed AI supercomputing clusters from three in 2018 to 44 in 2025. South Asia, Latin America and the Middle East and North Africa had corresponding totals of two, three and eight, according to the page. The source presents these figures as evidence of uneven infrastructure, but it does not describe the clusters’ capacity, ownership arrangements beyond the state-backed label, or performance.
The source also reports substantial regional differences in data-localization measures through 2024. East Asia and the Pacific had adopted 77 such measures, followed by sub-Saharan Africa with 71 and Europe and Central Asia with 66. North America had recorded three. Stanford HAI says this reflects different regional approaches to cross-border data flows. The excerpt does not define every measure, explain whether the measures are legally binding in the same way, or assess their effects on companies, researchers or the public.
In the United States, the number of AI-related witnesses in congressional hearings rose from five in 2017 to 102 in 2025. Industry’s share increased from 13% to 37%, making it the largest witness group, while academia’s share fell to 15%. The chapter separately estimates approximately $20.4 billion in U.S. AI-related contracts and grants from 2013 through 2024, compared with $285.9 billion in U.S. private investment in 2025 alone. In Europe, it reports approximately $3.7 billion in public commitments over 2013–2024, including $1.6 billion from the United Kingdom, $505 million from Germany and $320 million from France. The comparison uses different time periods, a limitation the figures themselves make clear.
Read the primary source: hai.stanford.edu ↗
Why it matters
The findings portray AI governance as an expanding but fragmented policy project. Governments are pursuing different combinations of national strategies, domestic infrastructure, data controls, public spending and legislative oversight. The figures also show why policy capacity cannot be assessed by counting laws alone: the source describes plans, infrastructure and investment, but does not establish how effectively any measure operates or what outcomes it produces.
The report’s data undercuts a simple description of AI policy as either absent or widespread. National strategies, localization measures, public contracts, infrastructure programs and legislative hearings are all forms of government engagement identified by the source. They are not interchangeable with enforceable regulation, however. A strategy can set priorities, and a localization measure can govern data movement, without the excerpt showing how either changes AI systems or protects the public.
The candidate headline’s broad claim cannot be evaluated as written because the supplied page neither mentions Becerra nor offers a single global count of AI regulations. AI sovereignty matters because it connects policy ambition to physical capacity. A government may seek more control over domestic AI capabilities, but the source shows that state-backed supercomputing infrastructure is distributed unevenly. That disparity is a meaningful constraint on how independently countries can develop or operate advanced AI systems, although the source does not quantify the relationship between cluster counts and national capability.
It also does not say whether countries without large clusters rely on foreign infrastructure, shared facilities or other arrangements. The regional data-localization figures suggest that governments are making materially different choices about where AI-related data may be stored or processed. For organizations operating across borders, those differences could make compliance and infrastructure planning more complex; that is a practical implication, not an outcome measured in the supplied source. Stanford HAI does not report whether localization measures have improved privacy, strengthened domestic industries, increased costs or reduced research cooperation. Those unknowns are central to judging whether the policies are effective rather than merely numerous.
The investment figures clarify the scale of the public-private imbalance while requiring careful interpretation. The reported $20.4 billion in U.S. public contracts and grants covers 12 years, whereas the $285.9 billion private-investment figure covers 2025 alone. The numbers therefore do not constitute a like-for-like annual comparison, and they do not prove that private investment determines policy. They do show that the source sees public funding as modest relative to private-sector spending. The rise in industry witnesses also raises questions about representation and influence, but the page does not analyze the content or outcomes of the hearings.
What to watch next
The next questions are whether strategies in emerging economies move from development to implementation, whether uneven supercomputing capacity narrows or widens, and how data-localization policies affect cross-border AI activity. In the United States and Europe, future data will need to show whether public investment and congressional oversight are accelerating, and how those efforts compare when measured over the same periods.
A key indicator will be whether the national strategies now being adopted or developed produce concrete programs, budgets, regulatory instruments and accountability mechanisms. The source establishes expansion in strategies, especially among emerging economies, but does not report implementation rates, deadlines, enforcement or measurable public benefits. Future reporting should distinguish a published strategy from an operational policy and identify which institutions are responsible for carrying it out.
Infrastructure is another major area to monitor. Europe and Central Asia’s increase from three to 44 state-backed AI supercomputing clusters is substantial within the report’s time frame, while other regions remain at much lower reported totals. The next useful evidence would include computing capacity, access rules, geographic distribution, energy requirements and whether the facilities support domestic researchers and companies. None of those details appears in the supplied excerpt, so cluster counts alone cannot establish equal or independent AI capability.
Data governance may continue to diverge. Stanford HAI records 77 localization measures in East Asia and the Pacific, 71 in sub-Saharan Africa, 66 in Europe and Central Asia and three in North America through 2024. Future updates should clarify whether the number of measures is rising, what kinds of data they cover and how they affect cross-border research, public services and commercial deployment. The source identifies different approaches to data flows, but it does not establish whether any approach is more effective.
The United States and Europe also warrant closer measurement. In the United States, future hearing data could show whether industry remains the largest witness group and whether academic, civil-society or public-sector participation changes. Public-investment figures should be reported on comparable annual and regional bases before conclusions are drawn about spending priorities. The source says recent commitments accelerated in the United Kingdom and Germany: in 2024, the U.K. committed $454.4 million and Germany $206.6 million. It does not provide comparable 2025 or 2026 figures, implementation results or evidence of how those commitments changed AI outcomes.
