Stanford AI Index 2026: AI is advancing faster than we can manage
AI is advancing faster than ever, but Stanford's 2026 AI Index warns that safety, trust and governance aren't keeping up.
AI is speeding ahead. The guardrails are not. Stanford’s 2026 AI Index shows that artificial intelligence is advancing rapidly across capability, adoption and investment. At the same time, the systems designed to measure, govern and build trust around AI are struggling to keep pace.
AI capability is moving faster
Frontier AI models made significant gains in 2025, particularly on demanding coding and reasoning tasks. Stanford’s report highlights the rapid improvement on benchmarks such as SWE-bench Verified, where performance moved from around 60% to close to 100% within a year.
The problem is that benchmarks themselves are becoming outdated faster. New evaluations designed to challenge advanced models can reach high performance levels much sooner than expected, making it harder to use a single test as a reliable measure of progress.
This creates a growing gap between what AI systems can do and how quickly researchers can develop meaningful ways to evaluate them.
Adoption and investment are surging
AI adoption is moving beyond technology companies. According to the AI Index, 88% of organisations reported using AI in 2025, while 70% said they were using generative AI in at least one business function.
Consumer adoption has also been unusually fast.
Generative AI reached 53% adoption within three years, a pace faster than the personal computer and the internet. However, adoption differs significantly between countries. In the United States, for example, the reported adoption rate was 28.3%.
The economic value of these tools is growing as well. Estimated annual consumer surplus from AI in the US reached $172 billion by early 2026, up from $112 billion a year earlier. The median value per user also tripled during the period.
Investment is accelerating alongside adoption. Global corporate AI investment more than doubled in 2025, with private investment increasing by 127.5%. Generative AI investment more than tripled and accounted for nearly half of private AI investment.
The US remained the leading destination for private AI investment, attracting roughly 23 times China's total in 2025.
Governance and trust are struggling to keep up
While AI capability and spending are rising, responsible AI practices are not progressing at the same speed.
The AI Index records 362 documented AI-related incidents in 2025, up from 233 in 2024. Transparency also declined, with the average score on the Foundation Model Transparency Index falling from 58 in 2024 to 40 in 2025.
This creates a difficult situation for organisations adopting increasingly powerful systems. Many are establishing AI governance roles, but knowledge gaps, limited budgets and uncertainty around regulations remain major barriers.
Infrastructure presents another challenge. The US has the largest number of data centres, while much of the world's leading AI chip production is concentrated in Taiwan. This creates supply chain risks as demand for computing power continues to grow. Governments are also trying to respond.
National AI strategies are expanding, while countries are investing in public computing infrastructure and developing rules around data and AI sovereignty. In Europe and Central Asia, the number of state-backed AI supercomputing clusters increased from three in 2018 to 44 in 2025.
Public trust, however, remains uneven. Only 31% of respondents in the US said they trusted their government to regulate AI effectively.
What comes next for AI
The central message from Stanford's AI Index 2026 is straightforward: AI capability and adoption are accelerating faster than the systems designed to manage them.
The next phase of the AI race may therefore depend on more than building increasingly capable models. The organisations and countries that can combine AI progress with reliable evaluation, responsible deployment, skilled workers and public trust may be better positioned to benefit from the technology's growth.