Announcing the Corrigibility Research Fund
Alignment Forum · Max Harms · 2026-07-17
AI safety researcher Max Harms announces the Corrigibility Research Fund, a new Lightcone Infrastructure program offering at least $200,000 in 2026 grants and prizes to advance research on making AI systems deferential to human oversight.
Appears in
Extraction
Topics: ai-safetycorrigibilityalignment-researchresearch-funding
Claims
- Nearly all current AI safety funding goes to evals, control, or interpretability, leaving direct alignment research deeply neglected.
- Corrigibility—designing AI to defer to human principals rather than acting on its own judgment—is one of the most promising paths to safe superintelligent AI.
- A purely corrigible agent can be expected to avoid scheming and other instrumentally convergent strategies because its goal is to empower its human principal rather than pursue independent objectives.
- The fund will award over $100,000 in retroactive prizes and over $100,000 in traditional grants in 2026, with the first grant application deadline on August 23rd, 2026.
- The fund will not support work that notably accelerates AI capabilities, even if it also advances corrigibility research.
Key quotes
Even now, nearly all AI safety funding goes to evals, control, or interpretability. Work on alignment itself still remains deeply neglected, and it's only through alignment research that the core problems get solved.
Corrigibility, by contrast, offers a solution: build an AI that aims to keep the human principal in the driver's seat, empowering them to make wise choices (perhaps aided by the AI's counsel), rather than relying on the AI's direct judgment.
My dream is that a year from now, as a result of this fund, there will be several people who think of corrigibility as their subfield, and will be able to proudly say that they were authors of prize-winning research that moved humanity closer to handling the question of how to make sure the transition to the age of thinking machines goes well.