Finding The evidence establishes an intellectual/adaptation relationship, a historical invitation to collaborate directly, a MIRI recommendation, institutional seed funding and a separate ControlAI sponsorship. It does not establish that Yudkowsky writes most of the channel's AI-risk material, owns it, controls its editorial decisions, or personally finances it. "Anti-AI" is also not a defined, audited subset of its output. Six decisive primary sources follow.
1. Yudkowsky adaptations and a direct production invitation Primary: https://www.lesswrong.com/posts/NjtHt55nFbw3gehzY/announcing-miri-s-new-ceo-and-leadership-team Relevant operator comments: Writer, October 11-13, 2023. Direct comment URL: https://www.lesswrong.com/posts/NjtHt55nFbw3gehzY/announcing-miri-s-new-ceo-and-leadership-team?commentId=mbcGNee7xm6rBfDQP Original full page read; comment-specific URL errored. Dates also visible on GreaterWrong's mirror of the same thread.
The channel operator says he met Yudkowsky at Manifest, invited him into the production Discord, and proposed turning his contributions to its writers into succinct animated explanations. He offered the same opportunity to other safety organizations and MIRI staff. Gretta Duleba requested an invitation; he subsequently said it had been sent. This is evidence of an offered collaboration channel, not a completed commissioning contract.
The operator identifies The Power of Intelligence and Sorting Pebbles as unchanged adaptations; The Parable of the Dagger as lightly edited; and The Hidden Complexity of Wishes as adapted with Yudkowsky's approval. He separately attributes an extinction/century explainer mainly to Matthew Barnett and the grabby-aliens explainer to himself.
In this same thread Duleba describes public concern as a route toward restrictions on frontier development and GPU distribution. Attribute that policy rationale to MIRI, not automatically to every animation.
2. Current named writing and editorial staff Primary interview: https://inkhavenspotlight.substack.com/p/tides-are-weirder-than-you-think Published April 25, 2026; Eneasz Brodski publication, Ben Pace interviewing Alice Liu. Interview introduction and discussion following the question about joining Rational Animations.
Liu is introduced as lead scriptwriter. She says she won a 2022 writing contest, joined part-time in January 2023, and became full-time in February 2025. She estimates roughly twelve scripts, several still in production. She describes Robert Miles as narrator and deeply involved editor who challenges statements he disagrees with; the CEO and other part-time writers also participate. Justin Kuiper is named as involved. She identifies an infohazards video as her work.
This is a participant account of a staffed production process, not an independent payroll audit or a complete authorship census. It establishes other actual writers and editorial input; it cannot by itself determine the fraction of AI-risk scripts attributable to any one person.
3. ControlAI's named sponsorship and its terms Primary: https://blog.controlai.org/p/controlais-creator-outreach Andrea Miotti and Max Winga, September 8, 2026. Sections ControlAI's Work / Our Partnerships. Linked sponsored example: https://www.youtube.com/watch?v=0bnxF9YfyFI ("What happens if AI just keeps getting smarter?"). Sponsor page fully read; YouTube retrieval exposed only minimal metadata.
ControlAI explicitly names Rational Animations as a sponsored partner. Its account describes buying 60-90-second ad segments, agreeing topics in advance, and ordinarily requiring a nonbinding script review. It says creators control the main content, subject to conditions against endorsing AI companies and requiring a citation to one of two expert-risk statements. The ad's call to action directs viewers to contact lawmakers. No amount paid to Rational Animations is supplied.
This is the sponsor's public admission and description of its process, not an inspected individual contract. It supports a paid-sponsorship relationship and intended political mobilization; it does not establish who wrote the whole video, hidden sponsorship, MIRI payment, or Yudkowsky control.
4. Institutional seed funding with explicit audience-building rationale Primary fund-manager report: https://forum.effectivealtruism.org/posts/zAEC8BuLYdKmH54t7/ea-infrastructure-fund-may-2021-grant-recommendations Published June 3, 2021. Section YouTube channel Rational Animations ($30,000), under Buck Shlegeris's grant reports.
EA Infrastructure Fund reports a $30,000 award to the newly created channel. The report calls this the Q1 2021 round, states April 2021 payout, and also frames aggregate totals as conditional on grantees accepting full awards. Preserve that report-level qualification rather than calling this a bank receipt.
The rationale explicitly compares the channel with Robert Miles's: attracting awareness of AI safety, deepening engagement among people already exposed, and helping involved audiences understand developments. This establishes a funder's declared theory of impact. It predates the 2026 ASI-ban campaign; do not retroactively label it funding for that exact campaign.
5. MIRI recommends Rational Animations Primary: https://intelligence.org/take-action/ Current page inspected September 12, 2026; Careers and Donation Targets, plus communication advice.
MIRI lists Rational Animations among organizations recommended for careers/donations and as examples of work it wants more of. The list expressly does not imply full mutual agreement. It supports a public recommendation, not a MIRI grant or governance relationship.
The same page argues public awareness can create policy demand for an international off-switch. It also expressly calls for candor and honesty and warns against manipulative conduct. A claim that MIRI publicly licenses deception would misrepresent this source; whether conduct matches policy is a separate evidentiary question.
6. Rational Animations' own identity Primary: https://www.rationalanimations.com/about-us Undated current page inspected September 12, 2026.
The organization describes an international remote team of about forty, a particular focus on AI safety/alignment, and animation/writing services. Its partner-logo section claims work with Sentience Institute, Blue Dot Impact, Manifold Markets, GiveDirectly, Metaculus, Charter Cities Institute and Forethought Foundation. These are claimed work relationships; the page does not specify whether each was a sponsor, commissioner, collaborator or donor. Do not encode them all as funders.
Grant leads not reconciled for publication The legacy Good Ventures search index still displays video-animation-cost awards of $1,000,000 (May 2023) and $1,495,355 (February 2024): https://www.goodventures.org/our-portfolio/grants-database/page/109/?limit=all https://www.goodventures.org/our-portfolio/grants-database/page/33/?limit=10000 Opening now redirects to https://goodventures.org/our-giving/ without those entries. Third-party grantmaking.ai displays $2,095,355 for one grant; other commentary quotes a different 2024 total. These may reflect amendments, different grantmakers or stale records; not resolved here. Do not sum or interchange them. Legacy Open Philanthropy grant pages did not provide usable current records in this pass. The current Coefficient Giving home page explicitly identifies itself as the renamed Open Philanthropy, but that does not reconcile these transactions.
No direct MIRI-to-Rational-Animations cash grant was established. No complete ledger, video-by-video credit census or private arrangement was inspected. A useful article formulation is that the channel translates parts of Yudkowsky's intellectual corpus into popular video, has documented contact with MIRI's communications effort, and separately participates in ControlAI's sponsored advocacy. Majority authorship remains unproven.