# METR reports funding for AI safety work [Canonical reading](https://escapemeridian.com/readings/metr-reports-funding-for-ai-safety-work-62756d4a) METR’s funding update is dated August 14, 2026. METR says it raised commitments of around $71 million over the previous six months for ambitious projects: studying autonomous capabilities, tracking recursive self-improvement, evaluating monitoring systems, conducting risk assessments and investigating AI incidents. It thanks named supporters, including The Audacious Project, foundations and individuals, and invites readers to donate. METR says it works to maintain independence from frontier AI companies, especially as its risk assessments become more consequential. It says it has not accepted funding from those companies or donations made by or at the direction of their staff. At the same time, METR says frontier AI companies provide a significant amount of free tokens for its evaluations, research and engineering. The update does not say how much those tokens contribute. METR says it is significantly expanding its team and starting new projects, and invites people to join. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source METR · Article Original source date: Aug 14, 2026 (2026-08-14T07:00:00.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-11-06T05:58:23.570Z. First observed: 2026-10-18T02:45:41.069Z. Retrieved: 2026-10-23T23:07:40.341Z. - [Funding update](https://metr.org/blog/2026-08-14-funding-update/) ## Limits Text as retrieved on October 23, 2026; originally published August 14, 2026. ## Open question How much do frontier AI companies’ free tokens contribute to METR’s evaluations, research and engineering? ## Source-stated METR says it raised commitments of around $71 million over the preceding six months. > In the last 6 months, METR raised commitments of around $71 million. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source-stated METR says the commitments will fund projects including studies of autonomous capabilities and recursive self-improvement, and work on monitoring systems, risk assessments and AI incidents. > This will fund ambitious projects: studying autonomous capabilities, tracking recursive self-improvement, evaluating monitoring systems, conducting risk assessments, investigating AI incidents, and more. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source-stated METR says it has not accepted funding from frontier AI companies or donations made by or at their direction, while those companies provide free tokens for its work. > We have not accepted funding from these companies, and we do not accept donations made by or at the direction of their staff. However, frontier AI companies currently provide a significant amount of free tokens for our evaluations, research, and engineering. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source-stated METR says it is significantly expanding its team and starting new projects. > We are significantly expanding our team and starting new ambitious projects. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source and contributor context METR · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Recursive self-improvement An AI system improving its own capabilities, such as its code, training methods or reasoning, then using the gains to improve further. Each cycle could hypothetically produce a more capable system. - [Definition reference](https://hai.stanford.edu/what-is-recursive-self-improvement) ## Frontier model One of the most capable AI models available at a given time; the set shifts as models improve. - [Definition reference](https://www.gov.uk/government/publications/frontier-ai-capabilities-and-risks-discussion-paper) ## METR METR studies what AI systems can do and how reliably they complete tasks, including work with human developers. - [Explore this context](/entities/metr) - [Reference](https://metr.org/) ## TIME TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services. - [Explore this context](/entities/time) - [Reference](https://time.com/) ## Editorial provenance Version 1. Publication ID: 376992d8-ceb1-bae4-df94-a38b89bfa896. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-06T05:54:15.370Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 6003fc3db0e4213a78852e60b5c069609fe9f346df8aefaedff6ade9e3097e1f. Source text hash: da5e5c829d319a29460832c0f9a53c1190bb92f71440717695bbb47323681ce8. ## Revision and source-check history ### Version 1 — 2026-11-06T05:58:23.570Z METR’s funding update is dated August 14, 2026. METR says it raised commitments of around $71 million over the previous six months for ambitious projects: studying autonomous capabilities, tracking recursive self-improvement, evaluating monitoring systems, conducting risk assessments and investigating AI incidents. It thanks named supporters, including The Audacious Project, foundations and individuals, and invites readers to donate. METR says it works to maintain independence from frontier AI companies, especially as its risk assessments become more consequential. It says it has not accepted funding from those companies or donations made by or at the direction of their staff. At the same time, METR says frontier AI companies provide a significant amount of free tokens for its evaluations, research and engineering. The update does not say how much those tokens contribute. METR says it is significantly expanding its team and starting new projects, and invites people to join. ## Original source METR · Article Original source date: Aug 14, 2026 (2026-08-14T07:00:00.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-11-06T05:58:23.570Z. First observed: 2026-10-18T02:45:41.069Z. Retrieved: 2026-10-23T23:07:40.341Z. - [Funding update](https://metr.org/blog/2026-08-14-funding-update/) ## Limits Text as retrieved on October 23, 2026; originally published August 14, 2026. ## Open question How much do frontier AI companies’ free tokens contribute to METR’s evaluations, research and engineering? ## Source-stated METR says it raised commitments of around $71 million over the preceding six months. > In the last 6 months, METR raised commitments of around $71 million. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source-stated METR says the commitments will fund projects including studies of autonomous capabilities and recursive self-improvement, and work on monitoring systems, risk assessments and AI incidents. > This will fund ambitious projects: studying autonomous capabilities, tracking recursive self-improvement, evaluating monitoring systems, conducting risk assessments, investigating AI incidents, and more. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source-stated METR says it has not accepted funding from frontier AI companies or donations made by or at their direction, while those companies provide free tokens for its work. > We have not accepted funding from these companies, and we do not accept donations made by or at the direction of their staff. However, frontier AI companies currently provide a significant amount of free tokens for our evaluations, research, and engineering. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source-stated METR says it is significantly expanding its team and starting new projects. > We are significantly expanding our team and starting new ambitious projects. - [Supporting source](https://metr.org/blog/2026-08-14-funding-update/) ## Source and contributor context METR · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Recursive self-improvement An AI system improving its own capabilities, such as its code, training methods or reasoning, then using the gains to improve further. Each cycle could hypothetically produce a more capable system. - [Definition reference](https://hai.stanford.edu/what-is-recursive-self-improvement) ## Frontier model One of the most capable AI models available at a given time; the set shifts as models improve. - [Definition reference](https://www.gov.uk/government/publications/frontier-ai-capabilities-and-risks-discussion-paper) ## METR METR studies what AI systems can do and how reliably they complete tasks, including work with human developers. - [Explore this context](/entities/metr) - [Reference](https://metr.org/) ## TIME TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services. - [Explore this context](/entities/time) - [Reference](https://time.com/) ## Editorial provenance Version 1. Publication ID: 376992d8-ceb1-bae4-df94-a38b89bfa896. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-06T05:54:15.370Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 6003fc3db0e4213a78852e60b5c069609fe9f346df8aefaedff6ade9e3097e1f. Source text hash: da5e5c829d319a29460832c0f9a53c1190bb92f71440717695bbb47323681ce8. ### uncertain — 2026-11-06T05:54:15.370Z — anthropic/claude-opus-5.5 Publication: 376992d8-ceb1-bae4-df94-a38b89bfa896 attribution — pass: This is a single METR post in METR's own voice. The reading consistently attributes its statements to METR ("METR says") and does not present them as established fact. > In the last 6 months, METR raised commitments of around $71 million. summary — uncertain: The summary is faithful, but a general reader may not grasp two central points. First, the reading never says what METR is; the source does say it does risk assessments and evaluations of AI. Second, "free tokens" is never explained, yet it carries the independence tension and the question. Tokens means AI model usage given at no cost. The definitions cover frontier AI but not tokens. Smallest repair: add a brief plain gloss of tokens or a term-library tooltip, and a short descriptor of METR from the source. The title's "AI safety work" is a reasonable gloss of the listed projects. > frontier AI companies currently provide a significant amount of free tokens for our evaluations, research, and engineering. date — pass: originalPublished is 2026-08-14T07:00Z. Its UTC day matches the stated August 14, 2026, and the URL path agrees. The reading uses the source's relative six-month window without making it more precise. It contains no hindsight or later events. The evidence is a publisher page, so no archive disclosure is needed. support — uncertain: Most claims match the source closely. Claim 3, however, says METR has not accepted donations "made by or at their direction," which drops the source's condition that the bar applies to companies' staff. The summary states this correctly. Smallest repair: restore "made by or at the direction of their staff." The note that the update does not quantify the token contribution is a correctly identified gap. The question follows from the evidence and is not leading. > we do not accept donations made by or at the direction of their staff media — pass: No media was selected, and no verified media candidates were supplied. That is appropriate for this text-only update. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/62756d4aab2b8df5dca70aea44a9f05f837e6805693fc04db43763d0f637254a/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/metr-reports-funding-for-ai-safety-work-62756d4a.json)