Global AI regulation
From voluntary principles to disclosure thresholds, in four years.
Updated 22 August 2026 · 4 events
How we got here
The first wave of AI governance was voluntary and produced commitments that were difficult to audit. The second wave attached obligations to model scale.
Scale thresholds are crude, and everyone drafting them knows it. They survive because they are administrable, which is a different property from being correct.
The current bill keeps a disclosure architecture instead of a licensing regime — a lighter instrument that still creates an evidentiary record.
What changed
Obligation moved from intention to measurement. Once a threshold exists, the argument shifts from whether to regulate to where the line sits.
Key events
- Voluntary principles
- First binding framework
- Defeat of the research exemption
Key people
- Nadia Haddad
- Committee members
- Research institutions
Where things stand
Disclosure thresholds intact, plenary vote unscheduled, research exemption defeated by four votes.
What happens next
- 01Plenary timing
- 02Measurement methodology
- 03Whether other jurisdictions copy the threshold
More dossiers
European security architecture
How a settlement written for one era became the subject of a live negotiation.
5 events · Security · Europe
The chip controls
Five years of export policy, and the supply chain it produced.
5 events · Chips · AI
