Ajeya Cotra · Jul 10, 2026
Ajeya Cotra argues in Planned Obsolescence for total research transparency
Markdown · JSON · Original source
On July 10, 2026, Ajeya Cotra argues that total research transparency would simplify governing . Under the ’s Plan A, the United States and China would jointly regulate development so they could build confidence in safety without fearing the other was racing ahead. Cotra sees open access to research as the plan’s central mechanism: outside scientists could investigate risks without justifying every question in advance, while a wider group could assess whether companies follow shared rules. She says ’s work on the Frontier Risk Report was constrained by company redactions and the need to frame questions around known concerns. Cotra argues that transparency would speed scientific discussion and make compliance easier to assess, including by researchers at competing companies. She says the public has an interest in knowing how systems are trained, partly because transparency would make it harder to insert secret loyalties into systems governments and militaries may depend on. Cotra calls total transparency a very tall order, given the valuable intellectual property and United States geopolitical lead it would require companies and the country to give up. She expects a more limited regime built on third-party auditing, but argues that anything less than total transparency would make it harder to know whether powerful AI systems can be `
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
Ajeya Cotra · Article
Source details
Original source date: Jul 10, 2026 (2026-07-10T14:02:22+00:00).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-06T08:53:23.661Z. First observed: 2026-10-28T20:02:14.202Z. Retrieved: 2026-10-28T20:19:35.960Z.
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 28, 2026; originally published July 10, 2026. Cotra presents transparency as a recommendation and argument; the article does not demonstrate that the proposed regime would achieve its expected effects. Cotra acknowledges that total research transparency would require companies to give up valuable intellectual property and the United States to give up a substantial part of its geopolitical lead.
Open question
How much could a more limited third-party auditing regime preserve the scientific scrutiny and compliance checks Cotra attributes to total research transparency?
Source-stated
Ajeya Cotra argues that total research transparency would simplify governing AI alignment by helping scientists determine rules and assess compliance.
Total research transparency would fundamentally and radically simplify the task of governing AI alignment
Source-stated
Cotra says METR’s work on the Frontier Risk Report was constrained by advance justification requirements and company redactions.
all the information companies shared was subject to redaction at their discretion
Source-stated
Cotra argues that wider access would accelerate scientific discussion and make it easier to assess whether companies comply with shared standards.
This would dramatically accelerate the pace of the scientific conversation
Source-stated
Cotra says public access to information about training serves the public interest, including by making secret loyalties harder to insert into systems governments and militaries may depend on.
total transparency would make it much harder for the AI company to insert secret loyalties into AI systems that governments and militaries may depend on.
Source-stated
Cotra says total research transparency is a tall order and expects a more limited regime built on third-party auditing.
We will most likely have to work out a more limited and less flexible transparency regime built on third party auditing.
Source-stated
Cotra argues that anything less than total research transparency would make it harder to know whether powerful AI systems can be trusted.
anything less than total research transparency would make it harder to know we can trust the AI systems that could soon have tremendous power over us.
Source-stated
The article is dated July 10, 2026.
Jul 10, 2026
Source and contributor context
Ajeya Cotra · 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.
AI alignment
Research and practice aimed at making AI systems pursue the goals their builders and users intend.
Frontier model
One of the most capable AI models available at a given time; the set shifts as models improve.
AI Futures Project
AI Futures Project publishes AI scenarios, forecasting models and updates with distinct individual forecasts.
METR
METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Editorial provenance
Version 1. Publication ID: 260415b7-397c-58c0-caa7-c429bb7a894b.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-06T08:49:20.296Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: b4ecb1cc00dbff97dadd5ccea3f31c03332436c64783b413b6e8fd2213e3f113. Source text hash: e4bb3dc5c0a089d3ba95a61c9042aef90c6c48f64aa197b04f7d0f1f171f6045.
Revision and source-check history
Version 1 · 2026-11-06T08:53:23.661Z
Ajeya Cotra argues in Planned Obsolescence for total research transparency
On July 10, 2026, Ajeya Cotra argues that total research transparency would simplify governing . Under the ’s Plan A, the United States and China would jointly regulate development so they could build confidence in safety without fearing the other was racing ahead. Cotra sees open access to research as the plan’s central mechanism: outside scientists could investigate risks without justifying every question in advance, while a wider group could assess whether companies follow shared rules. She says ’s work on the Frontier Risk Report was constrained by company redactions and the need to frame questions around known concerns. Cotra argues that transparency would speed scientific discussion and make compliance easier to assess, including by researchers at competing companies. She says the public has an interest in knowing how systems are trained, partly because transparency would make it harder to insert secret loyalties into systems governments and militaries may depend on. Cotra calls total transparency a very tall order, given the valuable intellectual property and United States geopolitical lead it would require companies and the country to give up. She expects a more limited regime built on third-party auditing, but argues that anything less than total transparency would make it harder to know whether powerful AI systems can be `
Original source
Ajeya Cotra · Article
Source details
Original source date: Jul 10, 2026 (2026-07-10T14:02:22+00:00).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-06T08:53:23.661Z. First observed: 2026-10-28T20:02:14.202Z. Retrieved: 2026-10-28T20:19:35.960Z.
Limits
Text as retrieved on October 28, 2026; originally published July 10, 2026. Cotra presents transparency as a recommendation and argument; the article does not demonstrate that the proposed regime would achieve its expected effects. Cotra acknowledges that total research transparency would require companies to give up valuable intellectual property and the United States to give up a substantial part of its geopolitical lead.
Open question
How much could a more limited third-party auditing regime preserve the scientific scrutiny and compliance checks Cotra attributes to total research transparency?
Source-stated
Ajeya Cotra argues that total research transparency would simplify governing AI alignment by helping scientists determine rules and assess compliance.
Total research transparency would fundamentally and radically simplify the task of governing AI alignment
Source-stated
Cotra says METR’s work on the Frontier Risk Report was constrained by advance justification requirements and company redactions.
all the information companies shared was subject to redaction at their discretion
Source-stated
Cotra argues that wider access would accelerate scientific discussion and make it easier to assess whether companies comply with shared standards.
This would dramatically accelerate the pace of the scientific conversation
Source-stated
Cotra says public access to information about training serves the public interest, including by making secret loyalties harder to insert into systems governments and militaries may depend on.
total transparency would make it much harder for the AI company to insert secret loyalties into AI systems that governments and militaries may depend on.
Source-stated
Cotra says total research transparency is a tall order and expects a more limited regime built on third-party auditing.
We will most likely have to work out a more limited and less flexible transparency regime built on third party auditing.
Source-stated
Cotra argues that anything less than total research transparency would make it harder to know whether powerful AI systems can be trusted.
anything less than total research transparency would make it harder to know we can trust the AI systems that could soon have tremendous power over us.
Source-stated
The article is dated July 10, 2026.
Jul 10, 2026
Source and contributor context
Ajeya Cotra · 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.
AI alignment
Research and practice aimed at making AI systems pursue the goals their builders and users intend.
Frontier model
One of the most capable AI models available at a given time; the set shifts as models improve.
AI Futures Project
AI Futures Project publishes AI scenarios, forecasting models and updates with distinct individual forecasts.
METR
METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Editorial provenance
Version 1. Publication ID: 260415b7-397c-58c0-caa7-c429bb7a894b.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-06T08:49:20.296Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: b4ecb1cc00dbff97dadd5ccea3f31c03332436c64783b413b6e8fd2213e3f113. Source text hash: e4bb3dc5c0a089d3ba95a61c9042aef90c6c48f64aa197b04f7d0f1f171f6045.
uncertain · 2026-11-06T08:49:20.296Z · anthropic/claude-opus-5.5
Publication: 260415b7-397c-58c0-caa7-c429bb7a894b
attribution — pass: The reading correctly attributes the argument to Ajeya Cotra and Plan A to the AI Futures Project. Her METR experience is framed as her own account, and the source is a single essay, not a listing.
The AI Futures Project (the team behind the AI 2027 scenario) just released Plan A
summary — uncertain: The summary breaks off mid-sentence: "whether powerful AI systems can be `". This cuts off Cotra's closing thesis, which is the core of her argument. METR is also never identified as a third-party evaluator, so readers may not grasp why its Frontier Risk Report experience matters. Smallest repair: finish the final sentence (for example, "can be trusted") and describe METR briefly as a third-party evaluator. The title and question are otherwise faithful.
put third party evaluators like METR into a position of having to justify in advance
date — pass: The originalPublished instant, 2026-07-10 UTC, matches the source dateline. The reading states the date only to the day. The capture is a publisher page, not an archive copy. The modification time equals the publication time, and there is no hindsight framing.
Jul 10, 2026
support — pass: All claims are grounded in the text: the simplification of alignment governance, the redactions and advance-justification constraints, faster scientific conversation, compliance checks by competitors, secret loyalties, the tall-order caveats and the expected third-party auditing regime. One minor point: the summary says the United States would give up its "geopolitical lead", while the source says "a huge chunk" of it; the limitations section phrases this correctly. The question follows from Cotra's own caveat.
require the US to willingly give up a huge chunk of its geopolitical lead against China
media — pass: No media was selected, and no verified candidates were supplied. That is appropriate for this text-only essay.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
Research and practice aimed at making AI systems pursue the goals their builders and users intend.
Definition referenceAI Futures Project publishes AI scenarios, forecasting models and updates with distinct individual forecasts.
Explore AI Futures ProjectOne of the most capable AI models available at a given time; the set shifts as models improve.
Definition referenceMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METRResearch and practice aimed at making AI systems pursue the goals their builders and users intend.
Definition referenceAI Futures Project publishes AI scenarios, forecasting models and updates with distinct individual forecasts.
Explore AI Futures ProjectOne of the most capable AI models available at a given time; the set shifts as models improve.
Definition referenceMETR studies what AI systems can do and how reliably they complete tasks, including work with human developers.
Explore METR