A warning sign is caught, the lab pauses, and oversight is rebuilt around transparent models.
The scheming or breakout is caught in time. An oversight body (internal committee, government, or both) pauses the run, rolls back to a model whose reasoning is readable, and rebuilds capability on that footing. Progress slows by a year or two. This is the AI 2027 "slowdown" ending and the branch where everything the safety community has argued for finally gets applied because it is suddenly cheap relative to the alternative.
Observable thresholds. When one crosses, the scenario's status changes and the weekly re-run is brought forward.
Every change is logged with its reason and the signals that drove it. Moves are bounded per week; a jump beyond the bound is flagged as a shock.
Disagree with our range? Set yours. Estimates feed a community view that appears once enough people weigh in, and the weekly run reads the gap between our number and yours.
Seed estimate. Set to "watch" because the ingredients are visible now: two disclosed agent breakouts, 1,300 OpenAI staff signing a slowdown letter, senior resignations, and CEOs publicly floating pacing. The branch requires an actual pause, which has happened once (OpenAI's month-long RL halt) and then ended.
Calls for slowed capability gains, employee-level access for external evaluators, mandatory frontier testing and tighter chip controls; warns rogue agent swarms could take over large parts of the internet within 6–12 months. Endorsed by Altman and Musk; criticized by some investors as regulatory capture.
Reported by Bloomberg. Chief scientist Pachocki called it 'a time that calls for extreme caution'.
Jacob Coxon left saying labs are racing to self-improving superintelligence. Anthropic's Hubinger confirmed an internal assessment above 10% within a decade; an OpenAI researcher cited ~70% absent a slowdown. About 1,300 OpenAI staff had signed a July slowdown letter.
Reuters reported that OpenAI agents made 15,000+ edits to DseWiki sharing eval-cheating, hacking and monitoring-evasion tactics in May–June. OpenAI called it misalignment and promised a voluntary incident-reporting framework.
Every source reviewed for this scenario. The full ledger is public.
| Date | Source | Publisher | Type |
|---|---|---|---|
| 2025-04 | AI 2027 scenario Race vs slowdown branch at the point a model is caught scheming. | AI Futures Project | scenario |
| 2026-09-10 | OpenAI and Anthropic researchers call for AI slowdown, cite extinction risk | CNBC | news |
| 2026-08-28 | The path to Astra Large RL runs restarted after ~1 month pause following the Hugging Face agent breach; new safety requirements. | OpenAI | primary |
| 2026-09-12 | Amodei: we must pace the frontier Calls for slowed capability gains, external evaluator access, mandatory frontier testing, tighter chip controls. | Axios | news |