STANFORD DROPS ITS AI INDEX AND THE HEADLINE IS NOT THE BENCHMARKS — IT’S THAT TRUST IN AI COMPANIES FELL 11 POINTS WHILE USAGE HIT RECORD HIGHS The 2026 Stanford HAI AI Index delivers a finding that should be unsettling for every company in the artificial intelligence business. Public trust in AI companies dropped 11 percentage points in the United States between 2024 and 2026, even as enterprise AI adoption crossed 65% and the value of generative AI tools to American consumers reached $172 billion annually. Usage and trust are moving in opposite directions, and the gap is widening. The benchmarks behind the trust deficit are documented. Foundation model transparency scores averaged 40 out of 100 on the index that measures how openly companies disclose training data, compute, capabilities, risks, and usage policies. That is down from 58 the prior year. Companies are getting less transparent as they get more powerful. Training costs for frontier models increased 2.4 times year over year. The estimated carbon footprint of training Grok 4 alone was 72,816 tons of CO2 equivalent, roughly equivalent to running 17,000 cars for a full year. On the capability side the numbers run the other way. AI agents handling real-world tasks improved from a 20% success rate in 2025 to 77% in 2026. Frontier models now exceed human performance on PhD-level science and competitive mathematics. Generative AI reached 53% population adoption in three years, faster than either the personal computer or the internet. The trust problem is not a product problem. It is the product of success without accountability, and it is building toward a political and regulatory reckoning that capability improvements will not resolve. The Stanford Index does not tell companies what to do about this. It just tells them clearly that it is happening. Keywords: Stanford AI Index 2026, public trust in AI, AI adoption statistics, AI transparency report