We measure how the world adopts AI.
Adoption is near universal. Measured outcomes are not. The Forum publishes the record: what has been deployed, what failed, who competes in each category, and what the rules now require.
Published
All researchEvery structural move points away from the billable hour
Accenture, read unit by unit: 28 disclosed moves, six vectors, and the one number that will settle the argument.
The model was not the problem.
A cancer centre spent USD 62.1 million on a decision support system that never reached a decision.
The six series
All researchThe Radar
One organisation, six branches, every disclosed move. What it owns, what it built, what it backed, what it depends on, and the layer it has failed to secure.
Open the series→Failures
One programme that did not work. The number, the source, and the structural cause reduced to a single failure mode.
Open the series→The Landscape
One category, defined and segmented on a single axis, with a maturity band on every box and a named buyer of record.
Open the series→Regulation
One obligation, jurisdiction or enforcement action, sourced to the instrument, the article and the date.
Open the series→Deployments
One large company running AI in production, opened on an object the reader touched this week.
Open the series→Capital
One investor mapped by round role rather than portfolio page. Thesis, layer exposure, concentration and gap.
Open the series→The Voice AI agents landscape
This week
The full calendarComing next
The calendarThe Voice AI agents landscape
EU AI Act Article 50 transparency, in force since 2 August 2026
Bühler Group
Andreessen Horowitz
The Radar: Nvidia
The Epic sepsis prediction model
The HR and talent landscape
California, everything binding right now
Who we are
The Global AI Forum is an independent research institution in San Francisco. We publish the record of how organisations and economies adopt artificial intelligence: what has been deployed, what failed, who competes in each category, and what the rules now require.
Two halves, kept apart on purpose. Research publishes six items a week, free and public, sourced on every claim. Advisory puts the same evidence to work for leadership teams through commissioned research, briefings and workshops.
Signals
All signalsAdoption is the wrong thing to measure
Use is near universal and tells you nothing. The question that matters is what stops working when you switch it off.
A category map cannot survive a participant funding it
One paid box does not corrupt that box. It corrupts every other box, because the reader now has to ask which ones were paid for.
The ceiling is arithmetic, not opinion
Decompose the work, find the share the system actually touches, and the maximum holds regardless of how good the model is.
Six pieces of research a week. One email on Sunday.
An organisation mapped, a failure documented, a category defined, an obligation explained, a company running AI in production, and an investor mapped by position. Free, and the archive is public.