· May 8, 2026
METR reviews Anthropic's automated R&D risk assessment
Markdown · JSON · Original source
published this review on May 8, 2026, assessing ’s argument that catastrophic risk from its models automating research and development is very low. METR says the report’s survey evidence has problems with sample size, question detail, framing, and calibrated responses, and that it miscounts a missing response as a negative one. METR also argues that the report overlooks the possibility that AI could substantially accelerate research and development before fully automating it. METR says evidence gathered after Opus 4.6’s release, including METR evaluations and the lack of public reports of the model automating a key domain, supports the report’s low-risk conclusion. But METR says the report’s own evidence is inadequate to establish that conclusion. It recommends changes to Anthropic’s surveys and more reporting of other evidence that could indicate AI progress. METR expects the public report to be updated and says it may publish another review if changes materially affect its opinions. The review also draws on non-public materials, so its assessment is not based only on public evidence.
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
· Article
Source details
Original source date: May 8, 2026 (2026-05-08T07:00:00.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-10-31T00:23:57.045Z. First observed: 2026-10-18T02:53:25.163Z. Retrieved: 2026-10-23T13:51:43.240Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 23, 2026; originally published May 8, 2026. says it used additional non-public materials, including information shared during a previous review; this summary cannot independently assess those materials. The supplied text is the executive summary of METR’s original review, not the full review documents.
Open question
Will changes to ’s public report or survey methods materially affect ’s assessment?
Source-stated
METR says the Anthropic report does not adequately support its conclusion that catastrophic risk from automated R&D is very low.
We do not think the report adequately supports its conclusion.
Source-stated
METR says survey design and interpretation problems limit what the cited survey results show about overall risk.
the cited results of the survey provide little evidence about the level of overall risk
Source-stated
METR says Anthropic miscounted a missing survey response as a negative response.
Anthropic summarizes the survey results in a way that miscounts one missing response to a question as a negative response.
Source-stated
METR says evidence after Claude Opus 4.6’s release supports the low-risk conclusion, while the report’s evidence alone is inadequate to establish it.
we agree with the bottom-line conclusion of the report — that the risk of a catastrophe from Opus 4.6 or a less capable Anthropic model automating R&D in any domain is very low — but we think the evidence presented in the report is inadequate to establish this.
Source-stated
METR recommends improvements to Anthropic’s internal model use surveys and reporting of other evidence that could indicate AI progress.
recommend improvements that Anthropic could make to their internal model use surveys
Source-stated
METR expects the public report to be updated and may publish a further review if changes materially affect its opinions.
if the updated public version includes any changes that materially affect our opinions, we will publish a further updated review.
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.
METR
METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Claude
Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Editorial provenance
Version 1. Publication ID: 66e9c0e1-1461-fbd1-a8c1-616aa73fed39.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-10-31T00:23:55.880Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_screen.
Observation: 553c25e0109f98fb2b01794f3e0293913412ca037c054f2138d7e576810bdc86. Source text hash: ae148d4abd6ee25541280ea4b1e694ee7939c857ebb3d85f2c6cef59af79736a.
Revision and source-check history
Version 1 · 2026-10-31T00:23:57.045Z
METR reviews Anthropic's automated R&D risk assessment
published this review on May 8, 2026, assessing ’s argument that catastrophic risk from its models automating research and development is very low. METR says the report’s survey evidence has problems with sample size, question detail, framing, and calibrated responses, and that it miscounts a missing response as a negative one. METR also argues that the report overlooks the possibility that AI could substantially accelerate research and development before fully automating it. METR says evidence gathered after Opus 4.6’s release, including METR evaluations and the lack of public reports of the model automating a key domain, supports the report’s low-risk conclusion. But METR says the report’s own evidence is inadequate to establish that conclusion. It recommends changes to Anthropic’s surveys and more reporting of other evidence that could indicate AI progress. METR expects the public report to be updated and says it may publish another review if changes materially affect its opinions. The review also draws on non-public materials, so its assessment is not based only on public evidence.
Original source
· Article
Source details
Original source date: May 8, 2026 (2026-05-08T07:00:00.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-10-31T00:23:57.045Z. First observed: 2026-10-18T02:53:25.163Z. Retrieved: 2026-10-23T13:51:43.240Z.
Limits
Text as retrieved on October 23, 2026; originally published May 8, 2026. says it used additional non-public materials, including information shared during a previous review; this summary cannot independently assess those materials. The supplied text is the executive summary of METR’s original review, not the full review documents.
Open question
Will changes to ’s public report or survey methods materially affect ’s assessment?
Source-stated
METR says the Anthropic report does not adequately support its conclusion that catastrophic risk from automated R&D is very low.
We do not think the report adequately supports its conclusion.
Source-stated
METR says survey design and interpretation problems limit what the cited survey results show about overall risk.
the cited results of the survey provide little evidence about the level of overall risk
Source-stated
METR says Anthropic miscounted a missing survey response as a negative response.
Anthropic summarizes the survey results in a way that miscounts one missing response to a question as a negative response.
Source-stated
METR says evidence after Claude Opus 4.6’s release supports the low-risk conclusion, while the report’s evidence alone is inadequate to establish it.
we agree with the bottom-line conclusion of the report — that the risk of a catastrophe from Opus 4.6 or a less capable Anthropic model automating R&D in any domain is very low — but we think the evidence presented in the report is inadequate to establish this.
Source-stated
METR recommends improvements to Anthropic’s internal model use surveys and reporting of other evidence that could indicate AI progress.
recommend improvements that Anthropic could make to their internal model use surveys
Source-stated
METR expects the public report to be updated and may publish a further review if changes materially affect its opinions.
if the updated public version includes any changes that materially affect our opinions, we will publish a further updated review.
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.
METR
METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Claude
Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Editorial provenance
Version 1. Publication ID: 66e9c0e1-1461-fbd1-a8c1-616aa73fed39.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-10-31T00:23:55.880Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_screen.
Observation: 553c25e0109f98fb2b01794f3e0293913412ca037c054f2138d7e576810bdc86. Source text hash: ae148d4abd6ee25541280ea4b1e694ee7939c857ebb3d85f2c6cef59af79736a.
passed · 2026-10-31T00:23:55.880Z · anthropic/claude-opus-5.5
Publication: 66e9c0e1-1461-fbd1-a8c1-616aa73fed39
attribution — pass: The reading consistently attributes the assessment, criticisms, agreement with the bottom line, and recommendations to METR. It also correctly identifies the reviewed document as Anthropic's February 2026 Risk Report section. This is a single review piece, not a listing.
This document is METR’s external review of the “Risks from automated R&D” section in the Anthropic Risk Report: February 2026
summary — pass: The reading leads with METR's central position: the report's conclusion is right, but its own evidence is inadequate. It spells out R&D in plain language and covers the survey critiques, the miscount, the overlooked acceleration-before-automation pathway, the recommendations and the update plan. It could have noted METR's stronger line that, on the report's evidence alone, it would likely disagree, but the core meaning is preserved. The question follows from the source's stated possibility of a further review.
If we had to solely rely on the evidence presented by Anthropic in the original Risk Report, we would likely disagree with the report’s conclusion
date — pass: evidence.originalPublished is 2026-05-08T07:00Z, whose UTC calendar day is May 8, 2026; the URL path matches. The reading states May 8, 2026 and does not present any later events. METR's expected report update is framed as an expectation, not an outcome. This is a publisher page, so no archive disclosure is required.
support — pass: Each claim is directly supported by the source text, with verbatim excerpts. The summary's mention of the acceleration-before-automation critique and the non-public materials is also source-stated. No unsupported benefits, costs or inferences are added.
the overall argument misses the possibility of substantial AI R&D acceleration before its full automation
media — pass: No media is selected and no verified media candidates were supplied, so nothing is misidentified or irrelevant.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicClaude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Explore ClaudeMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicClaude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Explore ClaudeMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METR