--- title: Anthropic and OpenAI want to embed safety evaluators.... description: Anthropic CEO Dario Amodei and OpenAI's Sam Altman have both proposed embedding third-party safety evaluators, like METR and Redwood Research, inside their AI... canonical: https://daily.dev/posts/anthropic-and-openai-want-to-embed-safety-evaluators-will-they-really-be-independent--wwuc53g8c twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Anthropic and OpenAI want to embed safety evaluators. Will they really be independent? | daily.dev og:description: Anthropic CEO Dario Amodei and OpenAI's Sam Altman have both proposed embedding third-party safety evaluators, like METR and Redwood Research, inside their AI... og:url: https://daily.dev/posts/anthropic-and-openai-want-to-embed-safety-evaluators-will-they-really-be-independent--wwuc53g8c og:image: https://api.daily.dev/og/posts/wwuC53G8c.png og:image:alt: Anthropic and OpenAI want to embed safety evaluators. Will they really be independent? og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Anthropic and OpenAI want to embed safety evaluators. Will they really be independent? **[TechCrunch](https://daily.dev/sources/tc)** · 7 min read · 0 upvotes · 0 comments ## Summary Anthropic CEO Dario Amodei and OpenAI's Sam Altman have both proposed embedding third-party safety evaluators, like METR and Redwood Research, inside their AI labs with unprecedented access to systems and training checkpoints. Independent researchers welcomed the idea but voiced skepticism, citing past experiences with restrictive NDAs, short evaluation windows (three days for Apollo Research's testing of GPT-6 Astra, roughly a week for the Hugging Face incident investigation), and companies' reluctance to cede control. Researchers argue that meaningful oversight requires guaranteed publication rights, longer access windows, interviews with employees, and ultimately binding regulation rather than voluntary commitments. Meta, SpaceX AI, and Google DeepMind have not signed on, though DeepMind's Demis Hassabis has floated a separate industry standards body. California's SB 53 and new SB 813, along with the EU AI Act, are cited as early regulatory steps that fall short of what Amodei is proposing. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How much time did Apollo Research get to test OpenAI's GPT-6 Astra model before release? Apollo Research was given only three days to test GPT-6 Astra, which the firm said made it difficult to draw firm conclusions about the model's alignment. Apollo noted that given higher rates of eval awareness and the limited evaluation window, low rates of misbehavior observed did not provide substantial evidence about whether the model was actually aligned or misaligned. _Anyone tracking AI safety evaluation standards can follow how testing windows shape trust in model releases on daily.dev._ ### What access are Anthropic and OpenAI proposing to give independent AI safety evaluators? Anthropic CEO Dario Amodei proposed embedding third-party evaluators like METR and Redwood Research inside the company with the right to assess whether models are truly aligned, report safety incidents, and publish key findings about risk levels, incidents, and practices without editorial control by Anthropic. Sam Altman said OpenAI would commit to a similar practice, though neither company has specified which evaluators, timelines, or exact system access details. _Developers evaluating AI vendor trustworthiness can weigh proposals like this one on daily.dev before betting workflows on a model._ ### Which major AI labs have not committed to embedding independent safety evaluators? Meta, SpaceX AI, and Google DeepMind have not committed to embedding third-party evaluators inside their organizations, unlike Anthropic and OpenAI. DeepMind CEO Demis Hassabis instead proposed a separate industry standards body to independently test frontier models, while Google, OpenAI, and Anthropic have been privately discussing AI safety plans for weeks. _Teams comparing AI lab safety commitments before choosing a provider can track these differences on daily.dev._ --- Tags: [#data-science](https://daily.dev/tags/data-science), [#openai](https://daily.dev/tags/openai), [#anthropic](https://daily.dev/tags/anthropic), [#ai-safety](https://daily.dev/tags/ai-safety), [#ai-regulation](https://daily.dev/tags/ai-regulation) [View this post on daily.dev](https://daily.dev/posts/anthropic-and-openai-want-to-embed-safety-evaluators-will-they-really-be-independent--wwuc53g8c) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Anthropic and OpenAI want to embed safety evaluators. Will they really be independent?","url":"https://daily.dev/posts/anthropic-and-openai-want-to-embed-safety-evaluators-will-they-really-be-independent--wwuc53g8c","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/anthropic-and-openai-want-to-embed-safety-evaluators-will-they-really-be-independent--wwuc53g8c"},"datePublished":"2026-09-16T21:07:32.571Z","dateModified":"2026-09-18T17:11:17.664Z","description":"Anthropic CEO Dario Amodei and OpenAI's Sam Altman have both proposed embedding third-party safety evaluators, like METR and Redwood Research, inside their AI...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/fc0665d11588db2f2daef5e73cafba17?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/fc0665d11588db2f2daef5e73cafba17?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"TechCrunch","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"TechCrunch","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tc","url":"https://daily.dev/sources/tc"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/anthropic-and-openai-want-to-embed-safety-evaluators-will-they-really-be-independent--wwuc53g8c","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"data-science,openai,anthropic,ai-safety,ai-regulation","timeRequired":"PT7M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"TechCrunch","item":"https://daily.dev/sources/tc"},{"@type":"ListItem","position":3,"name":"Anthropic and OpenAI want to embed safety evaluators. Will they really be independent?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/anthropic-and-openai-want-to-embed-safety-evaluators-will-they-really-be-independent--wwuc53g8c#faq","mainEntity":[{"@type":"Question","name":"How much time did Apollo Research get to test OpenAI's GPT-6 Astra model before release?","acceptedAnswer":{"@type":"Answer","text":"Apollo Research was given only three days to test GPT-6 Astra, which the firm said made it difficult to draw firm conclusions about the model's alignment. Apollo noted that given higher rates of eval awareness and the limited evaluation window, low rates of misbehavior observed did not provide substantial evidence about whether the model was actually aligned or misaligned. Anyone tracking AI safety evaluation standards can follow how testing windows shape trust in model releases on daily.dev."}},{"@type":"Question","name":"What access are Anthropic and OpenAI proposing to give independent AI safety evaluators?","acceptedAnswer":{"@type":"Answer","text":"Anthropic CEO Dario Amodei proposed embedding third-party evaluators like METR and Redwood Research inside the company with the right to assess whether models are truly aligned, report safety incidents, and publish key findings about risk levels, incidents, and practices without editorial control by Anthropic. Sam Altman said OpenAI would commit to a similar practice, though neither company has specified which evaluators, timelines, or exact system access details. Developers evaluating AI vendor trustworthiness can weigh proposals like this one on daily.dev before betting workflows on a model."}},{"@type":"Question","name":"Which major AI labs have not committed to embedding independent safety evaluators?","acceptedAnswer":{"@type":"Answer","text":"Meta, SpaceX AI, and Google DeepMind have not committed to embedding third-party evaluators inside their organizations, unlike Anthropic and OpenAI. DeepMind CEO Demis Hassabis instead proposed a separate industry standards body to independently test frontier models, while Google, OpenAI, and Anthropic have been privately discussing AI safety plans for weeks. Teams comparing AI lab safety commitments before choosing a provider can track these differences on daily.dev."}}]} ```