AIToday
AI Safety & AlignmentLessWrong AIPublished: Sep 4, 2026, 13:01 JST1 min read

Corrigibility Fund Manager Details Grant Process

Corrigibility Fund Manager Details Grant Process

Key takeaway

  • A new fund manager explained how they evaluate corrigibility grant applications.

  • They plan to give $50k to $150k.

  • All funds must benefit the public.

3 Key Points

  1. What happened

    The sole manager of the newly created Corrigibility Research Fund wrote up their grantmaking methods and experiences, aiming to disburse between $50k and $150k this round.

  2. Why it matters

    All funds must go to broad public benefit, which can include paying researchers for their time but excludes funding school, for-profit businesses, or political lobbying. The manager combines domain expertise with philanthropic micro-granting.

  3. What to watch

    The grant winners announcement is split into a separate post, so readers interested in outcomes should look for that.

Ask the AI about this article →

Context & Analysis

The Corrigibility Research Fund is a new initiative, and its sole manager is sharing their approach to grantmaking for the first time. The manager is a domain expert in alignment research, which gives them an advantage in evaluating proposals. The fund's constraint that all money must serve broad public benefit shapes what kinds of projects are eligible, excluding personal education or commercial ventures. This post focuses on methodology, not results, so the actual winners will be revealed separately. The manager hopes to inspire more work in corrigibility research by being transparent about their process.

FAQ

What is the funding range for this round?
The manager set out to disburse between $50k and $150k this round.
What types of projects cannot be funded?
Funds cannot go to schooling, for-profit businesses, or political lobbying. They must go to broad public benefit.

Get the latest AI Safety & Alignment news every morning

For example, today's edition would include:

  • Infosys and CrowdStrike partner on AI-discovered vulnerabilitiesSiliconANGLE AI · 2h ago
  • OpenAI's Astra model thinks beyond human oversightSemafor Tech · 2h ago
  • Instagram's AI labels misfire again, flagging real photosThe Verge AI · 2h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleOpenAI rolls out GPT-6 Astra, perfect on 3 benchmarks