Verification of AI agreements

Countries checking that each other keep AI deals, from compute limits to training halts.

Facts

Lane: Security and geopolitics
Wave: 8 of 12 (build order, not a date)
Needs: 17 boxes all the way back through hard links
Unlocks: 1 boxes downstream through hard links
Leverage: 0.2 points · stands in the way of 1 of 53 stages · rank 44 of 57
Cluster: Compute governance and geopolitics

Evidence as of October 2026

AI 2040 proposes inference-only verification, so the US and China could halt new frontier training while keeping deployed models. MIRI argues deals need chokepoints that can be monitored, such as fabs and concentrated datacentres.

01AI 2040
02MIRI: Refining MAIM

Around it

Needs · 4

International coordinationcapability · Emerging
Chip tracking and location verificationcapability · Emerging
Model weight securitycapability · Emerging
Audit and attributioncapability · Emerging · helps, not required

Unlocks · 1

Verified slowdownstage · Speculative

Loops and off-ramps

Capex pullbackoff-ramp · cut off
Taiwan chip disruptionoff-ramp · cut off
Power shortfalloff-ramp · cut off
Capability plateauoff-ramp · cut off
Training data ruled outoff-ramp · slowed

Openings researchers' judgement, October 2026 scouting

Startup room 46/100 · rank 43 of 124 Research room 66/100 · rank 13 of 124

Crowding2 Lab overlap2 Small-team fit3 Buyer today2 Research openness5

Depth: light
Players: MIRI TGT; Oxford AIGI; hardware-enabled mechanisms researchers
Wedge: Verifying that no large training run is happening under adversarial workload classification and distributed training. research