What the 2026 Stanford AI Index says about adoption, agents and reliability

A plain reading of the Index's published figures, then Solarascope's interpretation. The Index is not a Solarascope product.

1 September 2026 · 9 min read

The Stanford Institute for Human-Centered Artificial Intelligence published the 2026 AI Index in April 2026. Below are findings taken from the public Index pages, followed by Solarascope analysis. Citation is not endorsement.

Adoption

Source finding

Organisational AI adoption continued to rise in 2025, to 88% of surveyed organisations. Generative AI is now used in at least one business function at 70% of organisations. AI agent deployment was in the single digits across nearly all business functions. Generative AI reached 53% population-level adoption in three years.

Agents and reliability

Source finding

On OSWorld, which tests agents on real computer tasks, task success rose from about 12% to about 66% in a year. Agents still fail roughly one in three structured attempts. The Index also describes jagged performance: a system can win a gold medal at the International Mathematical Olympiad and still read analog clocks correctly only about half the time.

Incidents

Source finding

The Index records 362 documented AI incidents in 2025, up from 233 in 2024.

Solarascope analysis

The Index is not a brief for any vendor. It is a measurement of a widening gap: more use, more action, more incidents, and still-unreliable agents. That is the case for a reliance record, not for a slogan about responsible AI. Read the original chapters before you reuse any figure in a board pack.

Sources

Stanford HAI. AI Index Report 2026: Economy (April 2026)
Stanford HAI. AI Index Report 2026: Technical performance (April 2026)
Stanford HAI. AI Index Report 2026 (incidents) (April 2026)