“My funding decisions are formally recommendations.”In its own words: lesswrong.com
A Lightcone Infrastructure programme, announced in July 2026, that gives grants (typically $5k-$35k) and prizes for corrigibility research.
Show detailsHide details
Funds corrigibility research of any kind, from pure theory to empirical work and distillation, through grant rounds and retroactive prizes, with at least $200,000 available in 2026. Its fund manager recommends grants and Lightcone Infrastructure, which houses the fund, gives final approval.
Says it gives about $200k a year (2026)
Typical grant $6k ($6k–$20k)
Itemised here $201k in 19 grants
Money flows
Where its money came from and where it went. Pick years and what to show; hover a box or ribbon for details, click one to focus it, click the background to see everything.
How to apply
- Applications
- Closed
- Who can apply
- Individuals, Academics, New organisations, Organisations
- Where
- No formal eligibility rules; Lightcone Commons payment limits apply (Russia, India practically impossible). Grants must serve charitable purposes and cannot fund lobbying or political campaigns.
Both 2026 grant rounds have closed (23 Aug and 31 Oct 2026); no further round announced. Prizes need no application (at least $60k planned for mid-December 2026). Lightcone Commons itself takes rolling applications (next round closes mid-December), but the announcement doesn't say whether corrigibility grants continue there.
Biggest recipients
- Individual #5568$41k
- Individual #9498$34k
- Individual #4297$20k
- Individual #4816$20k
- Individual #6569$20k
- Individual #5783$13k
All grants
19 grants totalling $201k, newest first. Recipients link to their own pages; people who received grants without a public profile are shown by number.
2026 19 grants · $201k
| Date | Recipient | Amount | Purpose |
|---|---|---|---|
| Sep 2026 | Individual #5568 | $35k | Round 1 grant: behavioral benchmark of corrigibility and fine-tuning open-weight models toward it |
| Sep 2026 | Individual #4297 | $20k | Round 1 grant: activation-level geometry of corrigibility and related abstractions in LLMs |
| Sep 2026 | Individual #4816 | $20k | Round 1 grant: training architectures with a dedicated channel for models to flag their own misaligned behaviour |
| Sep 2026 | Individual #6569 | $20k | Round 1 grant: eight weeks formalising corrigibility in observable terms, with proofs verified in Lean |
| Sep 2026 | Individual #9498 | $20k | Round 1 grant: research group on corrigibility-centric alignment |
| Sep 2026 | Individual #9498 | $14k | Retroactive prize: "Corrigibility Transformation: Constructing Goals That Accept Updates" |
| Sep 2026 | Individual #5783 | $13k | Round 1 grant: AI-native games as environments for studying the principal-agent correction relationship |
| Sep 2026 | Individual #4893 | $9k | Retroactive prize: "Eval Cooperativeness May Be a Scalable Mitigation for Eval Gaming" |
| Sep 2026 | Carissa Cullen, Harry Garland, Individual #9126, Individual #46405, Christos Ziakas, Elliott Thornley | $6k | Retroactive prize: "Towards Shutdownable Agents: Generalizing Stochastic Choice in RL Agents and LLMs" source |
| Sep 2026 | Individual #6252 | $6k | Retroactive prize: "Assistance with CAST" |
| Sep 2026 | Individual #3302, Individual #2076, Samuel Marks, Owain Evans | $6k | Retroactive prize: "The Consciousness Cluster: Emergent preferences of Models that Claim to be Conscious" source |
| Sep 2026 | Individual #9318, Abishek Anand, Yu-Rou Tuan, Individual #5635, J. Zico Kolter, Aran Nayebi | $6k | Retroactive prize: "ROGUE: Misaligned Agent Behavior Arising from Ordinary Computer Use" source |
| Sep 2026 | Individual #5568 | $6k | Retroactive prize: "CAST Constitution, Empirical Work on Aspects of Corrigibility that are Unintuitive to LLMs, and other Preliminary Results" (unpublished) |
| Sep 2026 | Individual #7561 | $3k | Round 1 grant: fine-tuning an open-source model with corrigibility as its singular target |
| Sep 2026 | Individual #3214 | $3k | Retroactive prize: "Various Essays on Obedience" |
| Sep 2026 | Individual #1908 | $2k | Retroactive prize: "The corrigibility basin of attraction is a misleading gloss" |
| Sep 2026 | Individual #7561 | $2k | Retroactive prize: "A Structural Similarity Between Two Open Corrigibility Questions" and "Why Should Corrigible Agents Favor the Present?" |
| Sep 2026 | Individual #9351 | $2k | Round 1 grant: pilot study of simulated cases where an AI overrides a human decision for welfare reasons |
| Jul 2026 | Alex Turner | $9k | Retroactive prize (Round Zero): earlier corrigibility work; listed in the 30 Sep 2026 prizes post source |