AI Statistics 2026: Adoption, Investment, Business Use, Public Opinion, Jobs and Incidents
88% of organisations used AI in at least one business function in 2025, up from 55% in 2023, according to McKinsey survey data in the Stanford HAI AI Index 2026. Official surveys show lower rates. 19.95% of EU enterprises used AI in 2025 (Eurostat) and 19.8% of US businesses in spring 2026 (US Census Bureau). Global private AI investment reached $344.66 billion in 2025, and the United States took 82.9% of it, calculated by StatisticsGlobal. 49% of US adults now use AI chatbots, according to Pew Research Center.
Key statistics
- 88% of organisations used AI in at least one business function in 2025, according to McKinsey survey data in the Stanford HAI AI Index 2026.
- 19.95% of EU enterprises with 10 or more employees used AI technologies in 2025, up 6.47 percentage points on 2024, according to Eurostat.
- US private AI investment reached $285.9 billion in 2025, 23.1 times the $12.4 billion invested in China, according to the Stanford HAI AI Index 2026.
- 49% of US adults said they use AI chatbots such as ChatGPT, Gemini or Copilot in 2026, up from 33% in 2024, according to Pew Research Center.
- Reported AI incidents rose 55.4% in 2025, calculated by StatisticsGlobal from Stanford HAI data.
AI adoption statistics
How many people and organisations use AI depends a lot on who is asked and how. Business surveys of large firms show near universal use. Official surveys of all firms show about one in five.
88% of organisations used AI in at least one business function in 2025, up from 78% in 2024 and 55% in 2023. That is a rise of 33 percentage points in two years, calculated by StatisticsGlobal. The figures come from McKinsey's State of AI survey, as reported in the Stanford HAI AI Index 2026. Generative AI use followed the same path, reaching 79% of organisations.
Official statistics tell a different story. The US Census Bureau found that 19.8% of US businesses used AI in the two weeks before its survey period ending 3 May 2026. Even among firms with 250 or more staff, the rate was 37%. The gap with McKinsey is mostly about who is counted. McKinsey asks people at mostly larger organisations whether AI is used anywhere in the business. The Census Bureau asks every employer business, most of them small, about recent use in operations.
People are adopting AI faster than firms. 49% of US adults said they use AI chatbots in 2026, up from 33% in 2024, according to Pew Research Center. ChatGPT alone reached 44%, 2.4 times its 2023 level, calculated by StatisticsGlobal. Country rankings vary widely. The AI Index puts the United States 24th for generative AI adoption, at 28.3%, far behind the UAE at 64%. Use in the workplace is covered on our AI marketing statistics page, and chatbot brand shares on our AI chatbot market share page.
Source: McKinsey & Company State of AI surveys, as charted in Stanford HAI, AI Index 2026, Figure 4.3.2. Self-reported by survey respondents.
-
88% of organisations used AI in at least one business function in 2025, according to McKinsey survey data in the Stanford HAI AI Index 2026.
-
78% of organisations used AI in at least one business function in 2024, according to McKinsey survey data in the Stanford HAI AI Index 2026.
-
Organisational AI use rose 33 percentage points worldwide from 55% in 2023 to 88% in 2025, calculated by StatisticsGlobal from Stanford HAI data.
How we calculated it: 88% (2025) - 55% (2023) = 33 percentage points. Both values from Figure 4.3.2 of the AI Index 2026 Economy chapter.
-
79% of organisations regularly used generative AI in at least one business function in 2025, up from 71% in 2024, according to the Stanford HAI AI Index 2026.
-
Europe had the highest organisational AI use of any region in 2025, at 91%, ahead of North America at 90%, according to the Stanford HAI AI Index 2026.
-
Generative AI reached 53% population adoption within three years of launch, faster than the PC or the internet, according to the Stanford HAI AI Index 2026, with data to 2025.
-
The United States ranked 24th for generative AI adoption at 28.3% in 2025, behind the UAE at 64% and Singapore at 61%, according to the Stanford HAI AI Index 2026.
-
19.8% of US businesses used AI in the two weeks before the survey period ending 3 May 2026, according to the US Census Bureau.
-
37% of US firms with 250 or more employees used AI in their business operations in 2026, according to the US Census Bureau.
-
The US information sector had an AI use rate of 39.7% in 2026, about double the 19.8% national rate, according to the US Census Bureau.
-
US information firms used AI at 2.8 times the rate of retail firms in 2026, 39.7% against about 14%, calculated by StatisticsGlobal from US Census Bureau data.
How we calculated it: 39.7% (Information) / 14% (Retail Trade, "around 14%") = 2.84, rounded to 2.8. Collection period ending 3 May 2026.
-
49% of US adults said they use AI chatbots such as ChatGPT, Gemini or Copilot in 2026, up from 33% in 2024, according to Pew Research Center.
-
24% of US adults used AI chatbots at least daily in 2026, including 4% who use them almost constantly, according to Pew Research Center.
-
44% of US adults said they use ChatGPT in 2026, up from 34% in 2025 and 18% in 2023, according to Pew Research Center.
-
The share of US adults using ChatGPT was 2.4 times higher in 2026 than in 2023, calculated by StatisticsGlobal from Pew Research Center data.
How we calculated it: 44% (2026) / 18% (2023) = 2.44, rounded to 2.4. Pew changed the question wording in 2026, so treat the multiple as approximate.
-
ChatGPT reached 20 percentage points more US adults than Gemini in 2026, 44% against 24%, calculated by StatisticsGlobal from Pew Research Center data.
How we calculated it: 44% (ChatGPT) - 24% (Gemini) = 20 percentage points. Gemini was the second most used chatbot in the survey.
| Measure | Share | Period | Source |
|---|---|---|---|
| Organisations using AI in at least one function (worldwide) | 88% | 2025 | McKinsey via Stanford HAI |
| Organisations regularly using generative AI (worldwide) | 79% | 2025 | McKinsey via Stanford HAI |
| EU enterprises with 10+ employees using AI | 19.95% | 2025 | Eurostat |
| US businesses using AI in the past two weeks | 19.8% | Period ending 3 May 2026 | US Census Bureau |
| US firms with 250+ employees using AI | 37% | Spring 2026 | US Census Bureau |
| US adults who use AI chatbots | 49% | February 2026 | Pew Research Center |
| US adults who use ChatGPT | 44% | February 2026 | Pew Research Center |
AI investment by country
AI investment more than doubled in 2025. Most of the money went to US companies, and much of it to a small number of very large deals.
Global corporate AI investment reached $581.69 billion in 2025, up 129.9% on 2024, according to the Stanford HAI AI Index 2026. Private investment made up $344.66 billion of that total. Generative AI took $170.87 billion, or 49.6% of private AI investment, calculated by StatisticsGlobal.
The United States dominates. US private AI investment reached $285.9 billion, 23.1 times China's $12.4 billion. That is 82.9% of the world total and 13.7 times all of Europe, calculated by StatisticsGlobal. The United Kingdom came third at $5.9 billion. US investment grew 160.2% in a year, against 32.2% in China and 7.2% in Europe.
The money is concentrated in big rounds. 28 deals exceeded $1 billion in 2025, up from 15. The AI Index warns that private investment data likely understates China's total AI spending, because government guidance funds are not counted. US firms also made up 55.8% of newly funded AI companies, calculated by StatisticsGlobal. Data centre and chip spending is covered on our GPU market share and cloud market share pages.
Source: Stanford HAI, AI Index 2026, Figure 4.2.8 (Quid data).
-
Global corporate AI investment reached $581.69 billion in 2025, up 129.9% on 2024, according to the Stanford HAI AI Index 2026.
-
Global private investment in AI reached $344.66 billion in 2025, up 127.5% on 2024, according to the Stanford HAI AI Index 2026.
-
Global private investment in generative AI reached $170.87 billion in 2025, according to the Stanford HAI AI Index 2026.
-
US private AI investment reached $285.9 billion in 2025, 23.1 times the $12.4 billion invested in China, according to the Stanford HAI AI Index 2026.
-
US private AI investment was 13.7 times Europe's in 2025, $285.88 billion against $20.92 billion, calculated by StatisticsGlobal from Stanford HAI data.
How we calculated it: $285.88 billion (US) / $20.92 billion (Europe) = 13.67, rounded to 13.7. Both values from Figure 4.2.11 of the AI Index 2026 Economy chapter.
-
The United Kingdom had the third largest private AI investment in 2025, at $5.9 billion, according to the Stanford HAI AI Index 2026.
-
US private AI investment grew 160.2% in 2025, against 32.2% in China and 7.2% in Europe, according to the Stanford HAI AI Index 2026.
-
The United States had 1,953 newly funded AI companies in 2025, against 172 in the UK and 161 in China, according to the Stanford HAI AI Index 2026.
-
28 private AI investment events exceeded $1 billion in 2025, up from 15 in 2024, according to the Stanford HAI AI Index 2026.
-
Generative AI was worth an estimated $172 billion a year to US consumers by early 2026, up from $112 billion a year earlier, according to the Stanford HAI AI Index 2026.
| Rank | Country | Private AI investment 2025 | Newly funded AI companies 2025 |
|---|---|---|---|
| 1 | United States | $285.88 billion | 1,953 |
| 2 | China | $12.41 billion | 161 |
| 3 | United Kingdom | $5.90 billion | 172 |
| 4 | France | $4.36 billion | 84 |
| 5 | Canada | $4.28 billion | 79 |
| 6 | India | $4.09 billion | 108 |
| 7 | Germany | $3.89 billion | 92 |
| 8 | Israel | $3.58 billion | 64 |
| 9 | Australia | $2.52 billion | Not in top 12 |
| 10 | Saudi Arabia | $2.03 billion | Not in top 12 |
AI use by businesses in Europe
Eurostat runs the only official survey that compares AI use by firms across all EU countries. It covers enterprises with 10 or more employees.
19.95% of EU enterprises used AI in 2025, up 6.47 percentage points on 2024, according to Eurostat. That is 2.6 times the 2021 rate of 7.65%, calculated by StatisticsGlobal. Firm size matters most. 55.03% of large enterprises used AI, against 17% of small ones. Large firms use it at 3.2 times the rate of small firms.
The country gap is wider still. Denmark led at 42.03% and Romania was lowest at 5.21%, a ratio of 8.1 to 1, calculated by StatisticsGlobal. 13 of the 27 member states were above the EU average. Greece was the only country where use fell, from 9.81% to 8.93%. Estonia grew fastest since 2021, rising 8.4 times to 23.4%.
Text mining was the most used technology, at 11.75% of all enterprises. Among AI users, marketing or sales was the top purpose, at 34.70%. For firms that considered AI but did not use it, a lack of expertise was the main reason, cited by 70.89%. Legal uncertainty came second at 52.52%. Wider business data is on our small business statistics page.
- EU
- Denmark
- Germany
- Romania
Source: Eurostat, dataset isoc_eb_ai. No survey in 2022. Denmark's 2021 figure is higher than its 2023 figure in the published series.
-
19.95% of EU enterprises with 10 or more employees used AI technologies in 2025, up 6.47 percentage points on 2024, according to Eurostat.
-
The share of EU enterprises using AI was 2.6 times higher in 2025 than in 2021, calculated by StatisticsGlobal from Eurostat data.
How we calculated it: 19.95% (2025) / 7.65% (2021) = 2.61, rounded to 2.6. Eurostat dataset isoc_eb_ai, enterprises with 10 or more employees.
-
55.03% of large EU enterprises used AI in 2025, against 30.36% of medium and 17% of small enterprises, according to Eurostat.
-
Large EU enterprises used AI at 3.2 times the rate of small enterprises in 2025, calculated by StatisticsGlobal from Eurostat data.
How we calculated it: 55.03% (250+ employees) / 17% (10 to 49 employees) = 3.24, rounded to 3.2.
-
Denmark had the highest share of enterprises using AI in the EU in 2025, at 42.03%, followed by Finland at 37.82% and Sweden at 35.04%, according to Eurostat.
-
Romania had the lowest share of enterprises using AI in the EU in 2025, at 5.21%, followed by Poland at 8.36% and Bulgaria at 8.55%, according to Eurostat.
-
Danish enterprises used AI at 8.1 times the rate of Romanian enterprises in 2025, calculated by StatisticsGlobal from Eurostat data.
How we calculated it: 42.03% (Denmark) / 5.21% (Romania) = 8.07, rounded to 8.1. The gap is 36.82 percentage points.
-
13 of the 27 EU countries had an enterprise AI use rate above the EU average of 19.95% in 2025, calculated by StatisticsGlobal from Eurostat data.
How we calculated it: Count of EU member states with a 2025 value above 19.95% in dataset isoc_eb_ai: Denmark, Finland, Sweden, Belgium, Luxembourg, Netherlands, Austria, Germany, Estonia, Slovenia, Malta, Lithuania and Spain.
-
Greece was the only EU country where the share of enterprises using AI fell in 2025, from 9.81% to 8.93%, calculated by StatisticsGlobal from Eurostat data.
How we calculated it: Compared 2024 and 2025 values for all 27 member states in dataset isoc_eb_ai. Only Greece fell: 8.93% (2025) - 9.81% (2024) = -0.88 percentage points. Eurostat reports that 26 countries rose.
-
Estonia had the fastest growth in enterprise AI use in the EU from 2021 to 2025, rising 8.4 times from 2.77% to 23.4%, calculated by StatisticsGlobal from Eurostat data.
How we calculated it: 23.4% (2025) / 2.77% (2021) = 8.45 (8.448), rounded to 8.4. Ranked against the same ratio for all 27 member states; Lithuania was second at 4.8 times.
-
Text mining was the most used AI technology in EU enterprises in 2025, used by 11.75% of enterprises, according to Eurostat.
-
34.70% of EU enterprises using AI applied it to marketing or sales in 2025, the most common purpose, according to Eurostat.
-
70.89% of EU enterprises that considered but did not use AI cited a lack of relevant expertise in 2025, according to Eurostat.
| Rank | Country | 2025 | 2024 | Change, points |
|---|---|---|---|---|
| 1 | Denmark | 42.03% | 27.58% | +14.45 |
| 2 | Finland | 37.82% | 24.37% | +13.45 |
| 3 | Sweden | 35.04% | 25.09% | +9.95 |
| 4 | Belgium | 34.54% | 24.71% | +9.83 |
| 5 | Luxembourg | 33.61% | 23.73% | +9.88 |
| 6 | Netherlands | 33.21% | 23.06% | +10.15 |
| 7 | Austria | 29.95% | 20.27% | +9.68 |
| 8 | Germany | 25.97% | 19.75% | +6.22 |
| 9 | Estonia | 23.40% | 13.89% | +9.51 |
| 10 | Slovenia | 21.61% | 20.89% | +0.72 |
| 11 | Malta | 21.51% | 17.30% | +4.21 |
| 12 | Lithuania | 21.30% | 8.76% | +12.54 |
| 13 | Spain | 20.27% | 11.31% | +8.96 |
| EU average | 19.95% | 13.48% | +6.47 | |
| 14 | Ireland | 19.64% | 14.90% | +4.74 |
| 15 | France | 18.16% | 9.91% | +8.25 |
| 16 | Slovakia | 18.00% | 10.78% | +7.22 |
| 17 | Czechia | 17.60% | 11.26% | +6.34 |
| 18 | Italy | 16.40% | 8.20% | +8.20 |
| 19 | Croatia | 15.19% | 11.76% | +3.43 |
| 20 | Latvia | 12.21% | 8.83% | +3.38 |
| 21 | Portugal | 11.54% | 8.63% | +2.91 |
| 22 | Hungary | 10.37% | 7.41% | +2.96 |
| 23 | Cyprus | 9.27% | 7.90% | +1.37 |
| 24 | Greece | 8.93% | 9.81% | -0.88 |
| 25 | Bulgaria | 8.55% | 6.47% | +2.08 |
| 26 | Poland | 8.36% | 5.90% | +2.46 |
| 27 | Romania | 5.21% | 3.07% | +2.14 |
Public opinion on AI
More people use AI each year, but views of it are getting more cautious, especially in rich countries.
63% of US adults said AI is advancing too quickly in 2026, according to Pew Research Center. 40% expected a negative impact on society over the next 20 years, against 16% positive. Americans were 2.5 times as likely to expect harm as benefit, calculated by StatisticsGlobal. 71% said AI will make their personal information less secure.
Trust in regulators is low in the United States. 67% had little or no confidence in the government to regulate AI, up from 62% in 2024. The AI Index found the US had the lowest trust in its own government to regulate AI of any country surveyed, at 31%, against a 54% global average. Across 25 countries, a median of 53% trusted the EU to regulate AI, against 37% for the US and 27% for China.
Global views are mixed, not hostile. Across 37 countries in 2026, a median of 37% were more concerned than excited, while 41% were equally concerned and excited. The Ipsos data in the AI Index shows 59% saying AI products offer more benefits than drawbacks in 2025, up from 55%. Younger Americans have turned more wary. 55% of those aged 18 to 34 were more concerned than excited in 2026, up from 40% in 2024.
-
63% of US adults said AI is advancing too quickly in 2026, while 2% said too slowly, according to Pew Research Center.
-
40% of US adults expected AI to have a negative impact on society over the next 20 years in 2026, against 16% positive, according to Pew Research Center.
-
US adults were 2.5 times as likely to expect a negative as a positive impact of AI on society in 2026, calculated by StatisticsGlobal from Pew Research Center data.
How we calculated it: 40% (negative impact on society) / 16% (positive) = 2.5. 31% said equally positive and negative.
-
71% of US adults said AI will make their personal information less secure in 2026, against 3% more secure, according to Pew Research Center.
-
67% of US adults had little or no confidence in the US government to regulate AI effectively in 2026, up from 62% in 2024, according to Pew Research Center.
-
A median of 37% of adults across 37 countries were more concerned than excited about AI in 2026, while 41% were equally concerned and excited, according to Pew Research Center.
-
The share of US adults aged 18 to 34 more concerned than excited about AI rose from 40% in 2024 to 55% in 2026, according to Pew Research Center.
-
73% of US AI experts expected AI to improve how people do their jobs in 2024, against 23% of the US public, according to the Stanford HAI AI Index 2026.
-
US respondents had the lowest trust in their own government to regulate AI of any country surveyed in 2025, at 31%, against a 54% global average, according to the Stanford HAI AI Index 2026.
-
A median of 53% of adults across 25 countries trusted the EU to regulate AI effectively in 2025, against 37% for the US and 27% for China, according to the Stanford HAI AI Index 2026.
-
59% of people worldwide said AI products offer more benefits than drawbacks in 2025, up from 55% in 2024, according to the Stanford HAI AI Index 2026.
| Question | Answer | Share of US adults |
|---|---|---|
| Pace of AI advances | Too quickly | 63% |
| Impact on society over 20 years | Negative / positive | 40% / 16% |
| Impact on them personally over 20 years | Negative / positive | 31% / 23% |
| Effect on security of personal information | Less secure / more secure | 71% / 3% |
| Confidence in US government to regulate AI | Little or no confidence | 67% |
AI and jobs statistics
Most people expect AI to cost jobs. Hard evidence of job losses so far is narrower, and concentrated among young workers in exposed roles.
In 34 of 37 countries surveyed in 2026, people tended to expect AI to lead to fewer jobs, according to Pew Research Center. Fear is highest in rich countries. 76% said so in Australia and South Korea, and 71% in the United States, up from 64% in 2024. The median across all 37 countries was 46%, calculated by StatisticsGlobal. In 14 countries at least half of adults expected fewer jobs.
The evidence so far is narrower. The AI Index reports that US employment for software developers aged 22 to 25 had fallen close to 20% from its 2022 peak by September 2025, while older developers kept growing. About one-third of organisations in the McKinsey survey expected AI to reduce their workforce in the coming year. The AI Index notes that large job losses have not yet shown up in overall employment data.
Experts are far more upbeat than the public. 73% of US AI experts expected AI to improve how people do their jobs, against 23% of the public, in surveys from 2024. Meanwhile AI is already a work tool. 38% of employed US adults used chatbots for work tasks in 2026, and 58% of employees worldwide used AI at work in 2025. More on work patterns is on our remote work statistics page.
Source: Pew Research Center, topline Q57, 2026.
-
71% of US adults expected AI to lead to fewer jobs over the next 20 years in June 2026, up from 64% in August 2024, according to Pew Research Center.
-
Australia and South Korea had the highest shares expecting AI to cause fewer jobs in 2026, at 76% each, according to Pew Research Center.
-
In 34 of 37 countries surveyed in 2026, people tended to expect AI to lead to fewer jobs rather than more, according to Pew Research Center.
-
A median of 55% of adults in 18 high-income countries expected AI to mean fewer jobs in 2026, against 36% in 18 middle-income countries, according to Pew Research Center.
-
At least half of adults expected AI to lead to fewer jobs in 14 of 37 countries surveyed in 2026, calculated by StatisticsGlobal from Pew Research Center data.
How we calculated it: Count of countries with 50% or more answering "fewer jobs" in Pew topline Q57: US, Canada, France, Germany, Greece, Hungary, Italy, Poland, Spain, UK, Australia, South Korea, Chile and Peru.
-
The median share of adults expecting AI to lead to fewer jobs across 37 countries was 46% in 2026, calculated by StatisticsGlobal from Pew Research Center data.
How we calculated it: Median of the 37 "fewer jobs" country values in Pew topline Q57 (US uses the June 2026 survey). Range: 26% (Philippines, Nigeria) to 76% (Australia, South Korea).
-
US employment for software developers aged 22 to 25 had fallen close to 20% from its 2022 peak by September 2025, according to the Stanford HAI AI Index 2026.
-
About one-third of organisations surveyed in 2025 expected AI to reduce their workforce in the coming year, according to the Stanford HAI AI Index 2026.
-
38% of employed US adults used AI chatbots for tasks at work in 2026, according to Pew Research Center.
-
58% of employees worldwide used AI at work regularly or semi-regularly in 2025, according to the Stanford HAI AI Index 2026.
-
73% of US AI experts expected AI to improve how people do their jobs in 2024, against 23% of the US public, according to the Stanford HAI AI Index 2026.
| Country | Fewer jobs | More jobs | Not sure |
|---|---|---|---|
| Australia | 76% | 4% | 9% |
| South Korea | 76% | 7% | 9% |
| United States (June 2026) | 71% | 5% | 14% |
| Greece | 68% | 4% | 13% |
| Canada | 61% | 6% | 23% |
| United Kingdom | 61% | 7% | 24% |
| France | 60% | 4% | 27% |
| Germany | 51% | 4% | 22% |
| Japan | 49% | 8% | 20% |
| Brazil | 42% | 13% | 34% |
| India | 29% | 21% | 35% |
| Nigeria | 26% | 27% | 30% |
AI incidents and safety statistics
Reported AI incidents keep rising, while AI companies have become less open about how their models are built.
The AI Incident Database recorded 362 AI incidents in 2025, up from 233 in 2024, according to the Stanford HAI AI Index 2026. That is a rise of 55.4%, calculated by StatisticsGlobal. The annual count stayed under 100 until 2022. The OECD's automated monitor, which scans news in many languages, counts far more. It logged a monthly peak of 435 incidents in January 2026.
Transparency went backwards. The average Foundation Model Transparency Index score fell from 58 in 2024 to 40 in 2025, a drop of 31%, calculated by StatisticsGlobal. Reliability is still uneven. On one new accuracy benchmark, hallucination rates across 26 top models ranged from 22% to 94%.
Companies are putting more rules in place. The share of businesses with no responsible AI policies fell from 24% to 11%. Knowledge gaps were the top barrier, cited by 59%, ahead of budget limits at 48% and regulatory uncertainty at 41%. Security incidents more broadly are covered on our cybersecurity statistics page.
-
The AI Incident Database recorded 362 AI incidents in 2025, up from 233 in 2024, according to the Stanford HAI AI Index 2026.
-
Reported AI incidents rose 55.4% in 2025, calculated by StatisticsGlobal from Stanford HAI data.
How we calculated it: 362 incidents (2025) / 233 incidents (2024) - 1 = 55.4%. AIID figures are revised as records are added.
-
The OECD AI Incidents and Hazards Monitor logged a monthly peak of 435 AI incidents in January 2026, according to the Stanford HAI AI Index 2026.
-
The average Foundation Model Transparency Index score fell to 40 in 2025, from 58 in 2024, according to the Stanford HAI AI Index 2026.
-
The average transparency score of major AI model developers fell 31% in 2025, calculated by StatisticsGlobal from Stanford HAI data.
How we calculated it: 40 (2025 average) / 58 (2024 average) - 1 = -31.0%.
-
Hallucination rates across 26 top AI models ranged from 22% to 94% on a new accuracy benchmark in 2025, according to the Stanford HAI AI Index 2026.
-
The share of businesses with no responsible AI policies fell from 24% to 11% in 2025, according to the Stanford HAI AI Index 2026.
-
Knowledge gaps were the top barrier to responsible AI in 2025, cited by 59% of organisations, ahead of budget limits at 48%, according to the Stanford HAI AI Index 2026.
| Measure | 2024 | 2025 | Change |
|---|---|---|---|
| AI incidents recorded (AI Incident Database) | 233 | 362 | +55.4% (our calculation) |
| Average Foundation Model Transparency Index score | 58 | 40 | -31% (our calculation) |
| Businesses with no responsible AI policies | 24% | 11% | -13 points |
How these sources measure AI
AI statistics vary a lot because each source measures a different group, in a different way. Read each number with its scope.
The Stanford HAI AI Index compiles data from many providers. Its organisational adoption figures come from McKinsey's State of AI survey, which asks respondents at mostly larger organisations. The AI Index itself calls these self-reported and directional. Investment data comes from Quid and counts announced deals, so a few very large rounds can swing totals. The AI Index is licensed CC BY-ND 4.0, so we quote its numbers but redraw its charts from the published values.
Eurostat and the US Census Bureau are official surveys with large random samples. Eurostat covers EU firms with 10 or more employees and leaves out finance and agriculture. The Census Bureau's BTOS covers all US employer businesses and asks about AI use in the last two weeks. That is why both show about one firm in five using AI, far below the McKinsey figure.
Pew Research Center figures are representative surveys of adults. The US chatbot question changed wording in 2026, so the jump from 33% in 2024 is partly a change in method. Pew's global jobs data covers 37 countries, with the US asked separately in June 2026. Incident counts depend on what gets reported in the news. The AI Incident Database is human-reviewed and leans to English-language media, while the OECD monitor is automated and casts a wider net.
Frequently asked questions
What percentage of companies use AI?
It depends on the survey. 88% of organisations used AI in at least one business function in 2025, according to McKinsey data in the Stanford HAI AI Index. Official surveys covering all firms show about one in five. Eurostat found 19.95% of EU enterprises used AI in 2025, and the US Census Bureau found 19.8% of US businesses in spring 2026.
Is AI adoption growing faster than last year?
Yes. EU enterprise use rose 6.47 percentage points in 2025, to 19.95%, according to Eurostat. Organisational use in the McKinsey survey rose from 78% in 2024 to 88% in 2025. US adult chatbot use reached 49% in 2026, up from 33% in 2024, according to Pew Research Center.
Which country invests the most in AI?
The United States, with $285.9 billion of private AI investment in 2025, according to the Stanford HAI AI Index. That is 23.1 times China's $12.4 billion and 82.9% of the world total, calculated by StatisticsGlobal. The United Kingdom was third with $5.9 billion.
Which EU country uses AI the most?
Denmark, where 42.03% of enterprises used AI in 2025, according to Eurostat. Finland (37.82%) and Sweden (35.04%) followed. Romania was lowest at 5.21%.
Will AI cause job losses?
Most people expect it to. In 34 of 37 countries surveyed by Pew Research Center in 2026, people tended to expect fewer jobs. Measured effects so far are narrower. US employment for software developers aged 22 to 25 had fallen close to 20% from its 2022 peak by September 2025, according to the Stanford HAI AI Index.
Why are AI incidents rising?
More AI is in use and more harms are being reported. The AI Incident Database recorded 362 incidents in 2025, up 55.4% on 2024, calculated by StatisticsGlobal from Stanford HAI data. Part of the rise reflects better tracking and media attention, so counts are not a direct measure of harm.
Sources
- Stanford HAI, The 2026 AI Index Report, Chapter 4: Economy, April 2026, accessed 29 October 2026. Link · Saved data copy · Licence: CC BY-ND 4.0
- Eurostat, Use of artificial intelligence in enterprises (Statistics Explained), Data extracted December 2025, accessed 29 October 2026. Link · Saved data copy
- Pew Research Center, Americans and AI 2026: Chatbots, Smart Devices and Views on Impact, 17 June 2026, accessed 29 October 2026. Link · Saved data copy
- Stanford HAI, The 2026 AI Index Report, Chapter 3: Responsible AI, April 2026, accessed 29 October 2026. Link · Saved data copy · Licence: CC BY-ND 4.0
- Stanford HAI, The 2026 AI Index Report, April 2026, accessed 29 October 2026. Link · Saved data copy · Licence: CC BY-ND 4.0
- US Census Bureau, Large Firms With at Least 20 Employees Biggest AI Users, 26 May 2026, accessed 29 October 2026. Link · Saved data copy
- Eurostat, Artificial intelligence by size class of enterprise (isoc_eb_ai), dataset updated 15 June 2026, accessed 29 October 2026. Link · Saved data copy
- Pew Research Center, Globally, More People Expect AI to Cause Job Loss Than Growth, 17 September 2026, accessed 29 October 2026. Link · Saved data copy
- Stanford HAI, The 2026 AI Index Report, Chapter 9: Public Opinion, April 2026, accessed 29 October 2026. Link · Saved data copy · Licence: CC BY-ND 4.0
- Pew Research Center, Global views of AI 2026, full report with topline (PDF), 17 September 2026, accessed 29 October 2026. Link · Saved data copy
Changes to this page
- : Page published with Stanford HAI AI Index 2026 (Economy, Responsible AI and Public Opinion chapters), Eurostat 2025 enterprise survey, US Census Bureau BTOS to May 2026 and Pew Research Center 2026 surveys.