Washington stalled. Everyone else legislated.
Chapter 6 of the AI Index 2025 tracks a year in which US federal lawmaking barely moved while state legislatures, regulatory agencies, and governments worldwide accelerated. Legislative attention rose everywhere, spending rose everywhere, and the gap between proposing and passing a law got wider. The numbers:
6.1 — The year in global AI policy
The AI Index Steering Committee's selection of 2024's most consequential policy events. Money, law, and safety infrastructure, arriving roughly in that order.
- Feb 21, 2024
Singapore commits $1 billion to AI over five years
In his budget speech, Deputy Prime Minister and Finance Minister Lawrence Wong announced that the government will allocate over $1 billion over the next five years to support AI computation, talent development, and industry growth.
- Mar 11, 2024
Abu Dhabi launches a $100 billion AI investment firm
Abu Dhabi established MGX Fund Management Limited, a state-owned investment firm specializing in AI technologies, with a target of managing $100 billion in assets — part of the UAE's push to position itself as a global leader in AI innovation.
- Mar 13, 2024
European Parliament passes the EU AI Act
The landmark EU AI Act, the first of its kind, passed three months after a provisional agreement. It introduces transparency and reporting obligations, risk-based regulation, and bans on applications including social scoring, human manipulation, and biometric categorization using sensitive characteristics. Most provisions take effect in 2026 after a two-year implementation period.
- Mar 17, 2024
India launches the IndiaAI Mission with $1.25 billion
The initiative aims to build more than 10,000 GPUs through public-private partnerships, develop a national nonpersonal data platform, and support homegrown AI models and deep-tech startups. It also prioritizes ethical AI governance and expanding AI labs beyond major cities. Two days earlier, India had dropped a plan to require government approval before launching new AI models, switching to self-regulation after backlash from entrepreneurs and investors.
- Mar 20, 2024
France fines Google €250 million over training data
France's competition watchdog, the Autorité de la Concurrence, fined Google €250 million for using French news content to train Bard, now Gemini, without notifying media companies — a breach of EU intellectual property rules that the regulator said prevented publishers from negotiating fair prices. Google accepted the settlement.
- Mar 21, 2024
UN General Assembly adopts an AI resolution
Backed by more than 120 member states including China and endorsed without a vote by all 193, the assembly adopted a US-led resolution promoting 'safe, secure, and trustworthy' AI. It is not legally binding, but it calls on stakeholders to keep AI systems compliant with human rights law and recognizes their role in accelerating the Sustainable Development Goals.
- Apr 7, 2024
Canada pledges CA$2.4 billion
Canada's 2024 federal budget featured a CA$2.4 billion package to 'secure Canada's AI advantage,' directed at research and developer infrastructure, AI startups, small-business productivity, support for workers affected by AI, and the creation of a new Canadian AI Safety Institute.
- May 21, 2024
The AI Seoul Summit builds a network of safety institutes
Cohosted by the UK and South Korea, attending countries reported the safety measures they had adopted under the previous year's Bletchley Declaration and signed a letter of intent to establish a collaborative network of AI safety institutes. Institutes were pledged in Japan, France, Germany, Italy, Singapore, South Korea, Australia, Canada, and the European Union — following the first two, in the US and UK, in November 2023. Ten days earlier the UK AI Safety Institute had released Inspect, the first government-backed AI safety testing platform published under an open-source license.
- May 27, 2024
China launches a $47.5 billion semiconductor fund
China created its largest-ever state-backed investment fund for chips — the third phase of the 'Big Fund' that has supported the industry since 2014, including investments in the country's two largest chip foundries. The launch came amid tightening US export controls on the semiconductors underpinning AI training hardware. The next day, the European Commission unveiled its AI Office, with over 140 staff across five units, to implement the AI Act and enforce standards for general-purpose models.
- Aug 2, 2024
The UK withdraws £1.3 billion in promised funding
The incoming Labour government canceled £1.3 billion promised for technology and AI projects, saying the previous government's commitments had been 'underfunded.' The cuts included £500 million for the AI Research Resource, which funds computing power, and £800 million for the University of Edinburgh's exascale supercomputer.
- Sep 29, 2024
California's governor vetoes the state's frontier AI safety bill
Governor Gavin Newsom vetoed a bill that would have mandated safety testing for frontier models before public release and let the state attorney general sue over AI-related harm. Given California's status as the world's fifth-largest economy and home to many leading AI companies, the bill could have set a national precedent — a Brussels effect in reverse. Newsom argued it imposed excessive standards; critics of the bill said it would stifle the open-weight ecosystem. Twelve days earlier he had signed three bills (AB 2655, AB 2839, AB 2355) targeting deepfake election content, one of which a federal judge temporarily enjoined on October 2 over a Kamala Harris deepfake case, citing vague definitions of 'harmful' depictions.
- Nov 8, 2024
Saudi Arabia announces Project Transcendence
A $100 billion AI initiative spearheaded by the Public Investment Fund, aimed at establishing the kingdom as a global tech hub. It includes a partnership with Alphabet worth between $5 billion and $10 billion to develop Arabic-language AI models, aligning with Vision 2030's goal of diversifying the economy beyond oil.
- Nov 25, 2024
The International Network of AI Safety Institutes convenes
The US Departments of Commerce and State cohosted the inaugural meeting in San Francisco, focused on managing synthetic content risks, testing foundation models, and risk assessments for advanced systems. The US serves as inaugural chair; initial members are Australia, Canada, the European Union, France, Japan, Kenya, the Republic of Korea, Singapore, and the United Kingdom. The network secured over $11 million in global research funding commitments. Eleven days earlier the EU AI Office had published the first of four drafts of its General-Purpose AI Code of Practice.
- Dec 2, 2024
The US tightens semiconductor export controls again
The Commerce Department's Bureau of Industry and Security announced new restrictions covering 24 types of semiconductor manufacturing equipment, three types of software tools, and further limitations, further curbing China's ability to produce advanced chips. Seventeen days later the UN Security Council convened on AI in military contexts, with Secretary-General António Guterres warning that AI's evolution is outpacing governance frameworks and calling for 'international guardrails' — against a backdrop of widespread autonomous drone and robot use in Ukraine.
6.2 — Proposing is easy. Passing is not.
The AI Index tracked legislation containing the term 'artificial intelligence' across 114 countries from 2016 to 2024. The picture is one of enormous legislative attention converting into a modest number of actual laws — except in US state capitols.
Of the 114 countries analyzed, just 39 have enacted at least one AI-related law, for a global total of 204 laws since 2016. Annual output rose from 30 laws in 2023 to 40 in 2024, making it the second-highest year on record after 2022 — and up from a single law in 2016. Russia led 2024 with seven laws, followed by Belgium and Portugal with five each and the United States with four. Cumulatively since 2016 the United States leads with 27, then Portugal and Russia at 20 apiece, Belgium at 18, and South Korea at 13.
The federal bottleneck
In the US Congress, proposed AI-related bills climbed from 171 in 2023 to 221 in 2024 — nearly tripling since 2022. Four passed. That ratio is the single clearest fact in the chapter: legislative interest is real, and legislative output is not. Attention is measurable elsewhere too. AI mentions in US congressional committee reports hit 136 in the 118th session (2023–24), the highest count since tracking began in 2001 and up 83.8% from the 117th.
The states did not wait
State legislatures produced 131 AI-related laws in 2024, more than double the 49 of 2023 and a long way from the single law of 2016. California led 2024 with 22, followed by Utah (12), Maryland (8), and Virginia (6). Cumulatively from 2016 to 2024, California has passed 42 — more than twice any other state — ahead of Maryland, Virginia, and Utah at 17 each, and Illinois, Massachusetts, and Washington at 11 each. The substance varies widely: California's AI Transparency Act requires large providers to offer free AI detection tools and permanent disclosures on AI-generated content with a $5,000 fine per violation; Colorado's AI Act builds consumer protections around high-risk systems, requiring transparency, appeal rights, and regular impact assessments; Massachusetts allocated $1.26 billion to IT modernization including $25 million to integrate AI into state government operations.
Deepfakes: the one thing everyone agrees on
- Before 2024, five states — California, Washington, Texas, Michigan, and Minnesota — had laws regulating AI deepfakes in elections. In 2024, 12 more states including Oregon, New Mexico, and New York introduced similar regulations, bringing the cumulative total of state-level election deepfake laws to 20.
- Laws against intimate-image deepfakes are far more widespread: 36 state-level laws have been enacted, with 25 states covering all individuals and five covering minors only. Wyoming and Ohio are the only states with no intimate deepfake regulation of any kind.
- The two curves separated sharply after 2022 — election deepfakes went from a niche concern to a legislative priority in a single election cycle, while intimate-imagery laws had a head start and kept climbing.
Regulators moved faster than legislators
Regulations — the detailed directives executive agencies craft to enforce legislation — are where US federal action actually happened. The AI Index searched the Federal Register, which covers more than 436 agencies, and found 59 AI-related regulations in 2024, more than double the 25 of 2023. They came from a record 42 unique departments, up from 21 in 2023 and 17 in 2022. The Department of Health and Human Services issued the most (14), followed by the Centers for Medicare and Medicaid Services and the Commerce Department with seven each. Their content ranges from an executive order restricting bulk sensitive personal data transactions with countries of concern, to a CFPB circular limiting algorithmic scoring in hiring and promotion decisions, to export controls on semiconductor manufacturing equipment.
6.3 — Where public AI money actually goes
The AI Index used NLP to identify AI-related government contracts and grants. What emerges is not a spending race so much as two entirely different theories of what public AI money is for.
From 2013 to 2023 the United States led with roughly $5.2 billion spread across 2,678 unique AI contracts. The United Kingdom followed at $568 million over 555 contracts, then Germany ($278 million, 409 contracts), France ($190 million), and Spain ($100 million). The UK, Germany, and France together account for 56% of European public AI investment. Per capita the ordering shifts: the US spent $1.58 million per 100,000 inhabitants over the decade, with Finland ($1.29 million) and Denmark ($1.27 million) close behind and the UK at $0.84 million.
Two theories of AI, side by side
The composition of that spending is the real finding. In the United States, the Department of Defense took 75.0% of AI contract value in 2023, followed by Veterans Affairs (6.8%) and the Treasury (5.3%). In Europe, defense accounted for 0.84%. European AI tenders went instead to general public services (64.1%), education (12.3%), and health (7.4%) — together about 84% of the total. Same technology, opposite institutional gravity.
Europe is closing the gap
- Europe's 2023 AI public investment was roughly 67 times its 2013 level, against a fifteenfold increase for the United States. Europe saw a 400% year-over-year jump in 2017 and another 200% spike in 2019 — the same year national AI strategy releases peaked worldwide.
- In 2023 the United States spent $831.0 million on AI contracts against Europe's $581.4 million, leaving a gap of $249.6 million. That disparity widened until 2020 and has narrowed every year since.
- The composition within Europe shifted too. In 2023 countries that historically ranked lower — Romania, Greece, Hungary, Poland — broke into the top 10, alongside the UK ($262.6 million), Spain ($49.6 million), and Germany ($49.6 million).
Contracts are the small half
Contracts undercount public AI spending badly. In 2023 the US government awarded about $830 million in AI-related public tenders — and $4.5 billion in AI-related grants. Across 2013 to 2023 the US disbursed roughly $19.7 billion through 18,399 AI grants, at a median of $247,530 and an average of $1.07 million each. Grant funding grew nearly nineteenfold over the decade, from $230 million in 2013 to $4.5 billion in 2023, with an average annual growth rate of 40% from 2014 to 2020. The AI Index cautions explicitly against country-to-country comparisons on this data: grant-level figures exist only for the United States, and comparable data from the EU and China could not be collected.
The chapter in five lines
Headline findings from Chapter 6 · Policy and Governance.
US states are leading the way on AI legislation amid slow progress at the federal level: one state law in 2016, 49 in 2023, 131 in 2024.
Across 75 geographic areas, AI mentions in legislative proceedings rose 21.3% in 2024 to 1,889 — more than ninefold growth since 2016.
In 2024, 59 US AI-related federal regulations were introduced — more than double 2023's 25 — from a record 42 unique agencies.
Canada announced CA$2.4 billion, China launched a $47.5 billion semiconductor fund, France committed €109 billion, India pledged $1.25 billion, and Saudi Arabia's Project Transcendence represents a $100 billion initiative.
In 2023 the US Department of Defense took 75% of American AI contract spending. In Europe, defense took 0.84%.
Read the full Policy and Governance chapter
Chapter 6 (sections 6.1–6.3) with every figure and citation is free from Stanford HAI. Or head back to the report highlights and eight-chapter overview.
Open the AI Index Report 2025 →