Stanford HAI published its ninth annual AI Index report on April 13, spanning 400+ pages and tracking AI across capability, investment, adoption, and policy. Global AI investment reached $581 billion in 2025 — more than double the prior year — while SWE-bench coding scores jumped from 60% to near 100% in a single year and accuracy on Humanity's Last Exam climbed from 8.8% to 38.3%. Governance transparency among frontier model developers dropped from 58 to 40 points on the Foundation Model Transparency Index, and only 31% of US respondents trust government AI regulation, a record low. The report also notes training a frontier model like Grok 4 now generates over 72,000 tons of carbon-equivalent emissions.