The Conjunctive Hurdle For this outcome to materialize, several strict conditions must align: a top-five AI developer must release a top-ten model exhibiting a measured capability gain, and credible public documentation must prove that neither training compute nor research headcount exceeded the model's predecessor. While achieving capability gains without increasing resources is theoretically plausible through algorithmic breakthroughs or intense automation, the necessity for simultaneous,
Aligned with expectations that labs will continue to massively scale resources even as automation improves, the probability of a flat-compute, flat-headcount capability gain remains low at 13%.
As of October 2026, the conditions for this event are currently met: METR's time-horizon series has been halted, and no comparable successor exists. METR officially noted on September 8, 2026, that its dashboard "is no longer actively updated," explicitly warning that measurements above 16 hours are unreliable with its current task suite Task-Completion Time Horizons of Frontier AI Models [Task-Completion Time Horizons of Frontier AI Models](https://metr.org/ti
Set against related time-horizon forecasts, this probability was raised from 39% to 45% to enforce consistency with the high likelihood that near-term measurements remain frozen.
Status Quo and Resolution Criteria No binding regime currently restricts which models may be used internally for AI research. Existing and pending regimes, such as the EU AI Act, California SB 53, and New York's RAISE Act, regulate models as products or require disclosure, neither of which satisfies the criteria ifp.org. A qualifying rule must treat models specifically as research instruments—for instance, by imposing a model-age floor, a capability cap on internal deployment, or a mandator
Considered alongside related forecasts, this was maintained at 20%; while labs are highly likely to cross their automation thresholds by 2031 , the low 16% chance of a voluntary halt leaves a modest opening for binding government intervention.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited