The ninth AI Index report from Stanford HAI compiles global metrics on AI research, performance, adoption, economics, policy, and public opinion through 2025. It highlights rapid generative AI uptake, gaps in governance and evaluation, new economic and labor estimates, and standalone chapters on AI in science and medicine.

• Generative AI adoption hit ~53% of the population in just three years, faster than PCs or the internet, with corporate investment more than doubling in 2025 and 88% of organizations now using AI tools.
• Benchmarks are breaking down as labs disclose less, tests saturate, and independent evaluations sometimes contradict developer claims.
• 2025 saw diverging global AI policy: EU AI Act bans took effect, the US leaned deregulatory, Japan/South Korea/Italy passed new laws, and most new national AI strategies came from developing countries emphasizing "AI sovereignty."
• AI in science and medicine moved from isolated assistance to running full experimental workflows and system-wide hospital deployments (ambient scribes, diagnostic tools, FDA approvals).