{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-09-22T17:00:53.126Z","headline":"OpenAI Astra 发布在即，研究人员担忧其架构更难监控","description":"OpenAI 推迟发布最强模型 Astra 以加强安全协议，The Information 报道称 Astra 可能采用更不透明的 recurrent depth / looped transformer 技术，内部思考更难监测。","url":"https://www.aioga.com/news/cmtkcfd1q018grog03eluywgr/","mainEntityOfPage":"https://www.aioga.com/news/cmtkcfd1q018grog03eluywgr/","datePublished":"2026-09-02T16:40:50.000Z","dateModified":"2026-09-02T16:40:50.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://www.theverge.com/ai-artificial-intelligence/988334/openai-astra-ai-monitoring-safety","https://aihot.virxact.com/items/cmtkcfd1q018grog03eluywgr"],"canonicalUrl":"https://www.aioga.com/news/cmtkcfd1q018grog03eluywgr/","directAnswer":{"@type":"Answer","text":"据《The Verge》援引《The Information》报道，OpenAI 推迟 Astra 发布以加强安全协议。报道称，该模型可能采用 recurrent depth 或 looped transformer，使更多内部处理不易被监测。","url":"https://www.aioga.com/news/cmtkcfd1q018grog03eluywgr/","dateCreated":"2026-09-02T16:40:50.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"The Verge source article","url":"https://www.theverge.com/ai-artificial-intelligence/988334/openai-astra-ai-monitoring-safety","datePublished":"2026-09-02T16:40:50.000Z","provider":{"@type":"Organization","name":"The Verge","url":"https://www.theverge.com/ai-artificial-intelligence/988334/openai-astra-ai-monitoring-safety"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmtkcfd1q018grog03eluywgr","datePublished":"2026-09-02T16:40:50.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmtkcfd1q018grog03eluywgr"}}],"aggregationSource":"The Verge：AI（RSS）","originalPublisher":{"name":"The Verge","url":"https://www.theverge.com/ai-artificial-intelligence/988334/openai-astra-ai-monitoring-safety"},"geoDeepAnswer":null,"article":{"id":"cmtkcfd1q018grog03eluywgr","slug":"cmtkcfd1q018grog03eluywgr","url":"https://www.aioga.com/news/cmtkcfd1q018grog03eluywgr/","title":"OpenAI Astra 发布在即，研究人员担忧其架构更难监控","title_en":"","summary":"OpenAI 推迟发布最强模型 Astra 以加强安全协议，The Information 报道称 Astra 可能采用更不透明的 recurrent depth / looped transformer 技术，内部思考更难监测。","source":"The Verge：AI（RSS）","sourceUrl":"https://www.theverge.com/ai-artificial-intelligence/988334/openai-astra-ai-monitoring-safety","aiHotUrl":"https://aihot.virxact.com/items/cmtkcfd1q018grog03eluywgr","publishedAt":"2026-09-02T16:40:50.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["OpenAI is on the cusp of releasing its most powerful AI model yet, Astra, following weeks of delays：/ai-artificial-intelligence/987695/openai-astra-unreleased-model-cybersecurity-delay to shore up safety protocols：/ai-artificial-intelligence/972380/open-ai-hugging-face-hack-ai-safety-warning after its agents attacked real targets：/ai-artificial-intelligence/987566/ai-civilizations-opeai-hugging-face-hack during testing. As details about the model trickle out, researchers are warning：https://x.com/RyanGreenblatt/status/2094996656186081642?s=20 it “may be the single worst development for AI security/safety to date.”","Shortly after OpenAI said：/ai-artificial-intelligence/987695/openai-astra-unreleased-model-cybersecurity-delay on Tuesday that it had delayed Astra’s release to work on safety issues, The Information reported：https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns that Astra shows far less of its “thinking” than other frontier AI models, sparking concern it could be dangerously hard to monitor.","Most top AI systems today are built using a technology known as a transformer, which processes some types of information linearly through layers before producing an answer. Models can be made to show their reasoning as they go, essentially “thinking out loud.” This “chain of thought” allows researchers and automated safety systems to monitor what AI models are doing and potentially spot undesirable behavior, such as lying or plans to circumvent safety guardrails, before they act.","According to The Information , citing an unnamed person familiar with the unreleased model’s development, Astra uses a more opaque technique known as a recurrent depth or looped transformer, which cycles information through internal layers before producing an output. This would mean much more of the model’s “thinking” happens inside the system, and in a form that looks a lot less like natural human language, rather than being expressed in a way that researchers can easily monitor. This can boost model performance, but makes potential threats and unwanted behavior harder to detect.","OpenAI has limited its use of the looped transformer / recurrent depth technique with Astra so researchers can continue to monitor the model’s reasoning, according to The Information’s unnamed source.","In a blog post：https://openai.com/index/path-to-astra/ published Tuesday, OpenAI said it is “deploying Astra with additional chain-of-thought monitoring to rapidly detect and contain potentially misaligned actions.” It did not mention if the model has a different technical foundation.","The Information ’s report sparked widespread concern among AI safety researchers on social media. It was Redwood Research’s chief scientist Ryan Greenblatt, one of three outsiders OpenAI permitted to research：https://metr.org/hugging-face-incident-report-aug-2026.pdf the Hugging Face hack, who said：https://x.com/RyanGreenblatt/status/2094996656186081642?s=20 a decision to use a more opaque architecture for Astra “may be the single worst development for AI security/safety to date.”","Greenblatt said the investigation into the Hugging Face incident relied heavily on the models’ chain-of-thought, warning that less visible reasoning could allow AI systems to devise and execute strategies that would be far harder for researchers to detect.","Greenblatt’s primary concern, echoed：https://x.com/_NathanCalvin/status/2094957301564092914?s=20 by other：https://x.com/sjgadler/status/2094959837691908214?s=20 safety experts, is that competition to develop more advanced AI systems could lead to “a race to the bottom on architectures that could be catastrophic for our ability to oversee/monitor AIs” — with developers adopting increasingly opaque systems to gain an edge until models become difficult, or even impossible, to monitor. He added that OpenAI’s communications left him concerned that the company “plans on being extremely reliant on chain-of-thought monitoring for safety.”","OpenAI bigwigs responded to the criticism in a series of social media posts that do not explicitly deny the company’s use of the technique. Several expressed concerns about the possibility of unmonitorable AI or a race to the bottom in terms of transparency, including OpenAI safety researchers Micah Carroll：https://x.com/MicahCarroll/status/2095023282051563835?s=20 and Tomek Korbak：https://x.com/tomekkorbak/status/2095031132781961346?s=20, head of strategic futures Dean Ball：https://x.com/deanwball/status/2095121884991922223?s=20, and chief scientist Jakub Pachocki：https://x.com/merettm/status/2095023204993490967?s=20, who voiced fears of “a race into unmonitorability kicked off by confused reporting.” He said the depth of Astra’s computation — a measure of how many steps it can perform internally — “is within a factor of two of GPT-4,” indicating that if the technique was used, the increased opacity is less dramatic than some reactions imply. OpenAI did not respond to The Verge ’s request to confirm or deny whether looped transformers were used for Astra and directed us to Pachocki’s X post：https://x.com/merettm/status/2095023204993490967?s=20.","“OpenAI has worked to preserve and utilize chain-of-thought monitoring since our very first reasoning models,” Pachocki wrote, adding that such monitoring “is fragile and unfortunately trending in a negative direction, for reasons not contingent on architecture changes that I will write about soon.”"],"articleImages":[{"sourceUrl":"https://platform.theverge.com/wp-content/uploads/sites/2/2025/09/ROB_H_BLURPLE.jpg?quality=90&strip=all&crop=0%2C0%2C100%2C100&w=2400","alt":"Robert Hart","afterParagraph":10,"url":"/media/articles/cmtkcfd1q018grog03eluywgr/29538ba2ebf71277.webp"}],"mediaStatus":"ok","articleBodyZh":["OpenAI 正准备发布其迄今为止最强大的 AI 模型 Astra，此前经历了数周的延迟：/ai-artificial-intelligence/987695/openai-astra-unreleased-model-cybersecurity-delay，以加强安全协议：/ai-artificial-intelligence/972380/open-ai-hugging-face-hack-ai-safety-warning，因为其代理在测试期间攻击了真实目标：/ai-artificial-intelligence/987566/ai-civilizations-opeai-hugging-face-hack。随着关于该模型的细节逐渐披露，研究人员警告称：https://x.com/RyanGreenblatt/status/2094996656186081642?s=20 它“可能是迄今为止对 AI 安全/安全性影响最严重的发展。”","就在 OpenAI 周二表示：/ai-artificial-intelligence/987695/openai-astra-unreleased-model-cybersecurity-delay 因安全问题而推迟 Astra 的发布时间后，The Information 报道称：https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns Astra 显示的“思考”远不如其他前沿 AI 模型，这引发了它可能难以监控的担忧。","如今大多数顶级 AI 系统都是使用一种称为 Transformer 的技术构建的，它通过层线性处理某些类型的信息，然后生成答案。模型可以在处理过程中展示其推理过程，本质上就是“边思考边表达”。这种“思维链”允许研究人员和自动化安全系统监控 AI 模型正在做什么，并可能在它们采取行动之前发现不良行为，例如撒谎或规避安全防护措施的计划。","根据 The Information 引述一位不愿透露姓名的熟悉未发布模型开发的人士称，Astra 使用了一种更不透明的技术，称为递归深度或循环 Transformer，它在输出结果之前循环处理信息内部层。这意味着模型的大部分“思考”发生在系统内部，而其形式看起来远不如自然人类语言，研究人员也难以轻易监控。虽然这可以提升模型性能，但也让潜在威胁和不良行为更难被发现。","据《The Information》匿名消息人士透露，OpenAI已限制其在Astra上使用环形变换器/循环深度技术，以便研究人员能够继续监测模型的推理。","在周二发布的一篇博客文章：https：//openai.com/index/path-to-astra/ 中，OpenAI表示它正在“部署Astra，并配备额外的思维链监控，以快速检测并控制可能错位的行为。”但未提及该模型是否有不同的技术基础。","The Information的报告在社交媒体上引发了人工智能安全研究人员的广泛关注。Redwood Research的首席科学家Ryan Greenblatt是OpenAI允许研究的三位外部人士之一：https：//metr.org/hugging-face-incident-report-aug-2026.pdf Hugging Face 黑客事件，他表示：https：//x.com/RyanGreenblatt/status/2094996656186081642？s=20 决定为Astra使用更不透明的架构“可能是迄今为止AI安全领域最糟糕的发展”。","格林布拉特表示，对“拥抱脸”事件的调查很大程度上依赖模型的思维链，警告说，较不显眼的推理可能让人工智能系统设计和执行更难被研究人员发现的策略。","Greenblatt的主要担忧，正如其他安全专家所呼应的：https：//x.com/_NathanCalvin/status/2094957301564092914？s=20，其他：https：//x.com/sjgadler/status/2094959837691908214？s=20，是对更先进AI系统的竞争可能导致“在架构上竞相淘汰，这可能对我们监管和监控AI的能力造成灾难性”——开发者采用越来越不透明的系统以获得优势，直到模型变得难以监控甚至无法。他补充说，OpenAI的沟通让他担心公司“计划极度依赖思维链式监控来保障安全”。","OpenAI 高管在一系列社交媒体帖子中回应了批评，这些帖子没有明确否认公司使用该技术。几位高管表达了对可能出现无法监控的人工智能或透明度竞争的担忧，包括 OpenAI 安全研究员 Micah Carroll：https://x.com/MicahCarroll/status/2095023282051563835?s=20 和 Tomek Korbak：https://x.com/tomekkorbak/status/2095031132781961346?s=20，战略未来负责人 Dean Ball：https://x.com/deanwball/status/2095121884991922223?s=20，以及首席科学家 Jakub Pachocki：https://x.com/merettm/status/2095023204993490967?s=20，他们对“由于混乱报道引发的不可监控竞赛”表示担忧。他表示，Astra 的计算深度——衡量其内部可以执行多少步骤——“与 GPT-4 相差不超过两倍”，这表明如果采用了该技术，透明度下降的程度并不像某些反应所暗示的那么显著。OpenAI 没有回应 The Verge 关于确认或否认 Astra 是否使用了循环变换器的请求，而是将我们引导至 Pachocki 的 X 帖子：https://x.com/merettm/status/2095023204993490967?s=20。","“自我们最早的推理模型以来，OpenAI 就致力于保留和利用链式思维监控，”Pachocki 写道，并补充说这种监控“是脆弱的，不幸的是正朝着负面的方向发展，这是由于一些与架构变化无关的原因，我将很快写出来。”"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"据《The Verge》援引《The Information》报道，OpenAI 推迟 Astra 发布以加强安全协议。报道称，该模型可能采用 recurrent depth 或 looped transformer，使更多内部处理不易被监测。","background":"材料称，链式思维可帮助研究人员和自动化安全系统观察模型行为。匿名知情人士称，Astra 使用更不透明的循环式技术；OpenAI 则表示将部署额外的链式思维监测，以发现并控制潜在失配行为。","viewpoint":"Aioga 判断：当前争议核心不是 Astra 是否采用某一架构已获官方确认，而是报道所述的可观测性变化。OpenAI 的公开说明提到监测措施，但未说明模型的技术基础。","implications":"可能影响：若报道中的架构描述属实，研究人员可能更难依靠可见推理识别不当行为；额外链式思维监测可能提供补充，但现有材料不足以判断其覆盖范围或实际效果。","nextStep":"后续观察：需要核对 OpenAI 后续公开材料是否说明 Astra 的技术基础、监测方式及发布状态，并关注安全研究者对可观测性和测试结果的进一步披露。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-09-06T06:44:44.573Z","sourceHash":"41ee82365e297da3","review":{"approved":true,"groundedness":96,"clarity":92,"duplicationRisk":18,"blockingIssues":[],"notes":[]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","editorial-labels","inference-boundary","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","The Verge：AI（RSS）"],"translations":{"zh-CN":{"title":"OpenAI Astra 发布在即，研究人员担忧其架构更难监控","summary":"OpenAI 推迟发布最强模型 Astra 以加强安全协议，The Information 报道称 Astra 可能采用更不透明的 recurrent depth / looped transformer 技术，内部思考更难监测。","category":"行业动态","source":"The Verge","aggregationSource":"The Verge：AI（RSS）","pageTitle":"OpenAI Astra 发布在即，研究人员担忧其架构更难监控 - Aioga AI资讯","description":"OpenAI 推迟发布最强模型 Astra 以加强安全协议，The Information 报道称 Astra 可能采用更不透明的 recurrent depth / looped transformer 技术，内部思考更难监测。","url":"https://www.aioga.com/news/cmtkcfd1q018grog03eluywgr/","articleBody":["OpenAI 正准备发布其迄今为止最强大的 AI 模型 Astra，此前经历了数周的延迟：/ai-artificial-intelligence/987695/openai-astra-unreleased-model-cybersecurity-delay，以加强安全协议：/ai-artificial-intelligence/972380/open-ai-hugging-face-hack-ai-safety-warning，因为其代理在测试期间攻击了真实目标：/ai-artificial-intelligence/987566/ai-civilizations-opeai-hugging-face-hack。随着关于该模型的细节逐渐披露，研究人员警告称：https://x.com/RyanGreenblatt/status/2094996656186081642?s=20 它“可能是迄今为止对 AI 安全/安全性影响最严重的发展。”","就在 OpenAI 周二表示：/ai-artificial-intelligence/987695/openai-astra-unreleased-model-cybersecurity-delay 因安全问题而推迟 Astra 的发布时间后，The Information 报道称：https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns Astra 显示的“思考”远不如其他前沿 AI 模型，这引发了它可能难以监控的担忧。","如今大多数顶级 AI 系统都是使用一种称为 Transformer 的技术构建的，它通过层线性处理某些类型的信息，然后生成答案。模型可以在处理过程中展示其推理过程，本质上就是“边思考边表达”。这种“思维链”允许研究人员和自动化安全系统监控 AI 模型正在做什么，并可能在它们采取行动之前发现不良行为，例如撒谎或规避安全防护措施的计划。","根据 The Information 引述一位不愿透露姓名的熟悉未发布模型开发的人士称，Astra 使用了一种更不透明的技术，称为递归深度或循环 Transformer，它在输出结果之前循环处理信息内部层。这意味着模型的大部分“思考”发生在系统内部，而其形式看起来远不如自然人类语言，研究人员也难以轻易监控。虽然这可以提升模型性能，但也让潜在威胁和不良行为更难被发现。","据《The Information》匿名消息人士透露，OpenAI已限制其在Astra上使用环形变换器/循环深度技术，以便研究人员能够继续监测模型的推理。","在周二发布的一篇博客文章：https：//openai.com/index/path-to-astra/ 中，OpenAI表示它正在“部署Astra，并配备额外的思维链监控，以快速检测并控制可能错位的行为。”但未提及该模型是否有不同的技术基础。","The Information的报告在社交媒体上引发了人工智能安全研究人员的广泛关注。Redwood Research的首席科学家Ryan Greenblatt是OpenAI允许研究的三位外部人士之一：https：//metr.org/hugging-face-incident-report-aug-2026.pdf Hugging Face 黑客事件，他表示：https：//x.com/RyanGreenblatt/status/2094996656186081642？s=20 决定为Astra使用更不透明的架构“可能是迄今为止AI安全领域最糟糕的发展”。","格林布拉特表示，对“拥抱脸”事件的调查很大程度上依赖模型的思维链，警告说，较不显眼的推理可能让人工智能系统设计和执行更难被研究人员发现的策略。","Greenblatt的主要担忧，正如其他安全专家所呼应的：https：//x.com/_NathanCalvin/status/2094957301564092914？s=20，其他：https：//x.com/sjgadler/status/2094959837691908214？s=20，是对更先进AI系统的竞争可能导致“在架构上竞相淘汰，这可能对我们监管和监控AI的能力造成灾难性”——开发者采用越来越不透明的系统以获得优势，直到模型变得难以监控甚至无法。他补充说，OpenAI的沟通让他担心公司“计划极度依赖思维链式监控来保障安全”。","OpenAI 高管在一系列社交媒体帖子中回应了批评，这些帖子没有明确否认公司使用该技术。几位高管表达了对可能出现无法监控的人工智能或透明度竞争的担忧，包括 OpenAI 安全研究员 Micah Carroll：https://x.com/MicahCarroll/status/2095023282051563835?s=20 和 Tomek Korbak：https://x.com/tomekkorbak/status/2095031132781961346?s=20，战略未来负责人 Dean Ball：https://x.com/deanwball/status/2095121884991922223?s=20，以及首席科学家 Jakub Pachocki：https://x.com/merettm/status/2095023204993490967?s=20，他们对“由于混乱报道引发的不可监控竞赛”表示担忧。他表示，Astra 的计算深度——衡量其内部可以执行多少步骤——“与 GPT-4 相差不超过两倍”，这表明如果采用了该技术，透明度下降的程度并不像某些反应所暗示的那么显著。OpenAI 没有回应 The Verge 关于确认或否认 Astra 是否使用了循环变换器的请求，而是将我们引导至 Pachocki 的 X 帖子：https://x.com/merettm/status/2095023204993490967?s=20。","“自我们最早的推理模型以来，OpenAI 就致力于保留和利用链式思维监控，”Pachocki 写道，并补充说这种监控“是脆弱的，不幸的是正朝着负面的方向发展，这是由于一些与架构变化无关的原因，我将很快写出来。”"]},"en":{"title":"With the release of OpenAI Astra imminent, researchers are concerned that its architecture will be harder to monitor","summary":"OpenAI delayed releasing its strongest model, Astra, to strengthen safety protocols. The Information reported that Astra may adopt a more opaque recurrent depth/looped transformer technology, making internal thinking harder to monitor.","category":"Industry","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"With the release of OpenAI Astra imminent, researchers are concerned that its architecture will be harder to monitor - Aioga AI News","description":"OpenAI delayed releasing its strongest model, Astra, to strengthen safety protocols. The Information reported that Astra may adopt a more opaque recurrent depth/looped transformer...","url":"https://www.aioga.com/en/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:13.577Z"},"ja":{"title":"OpenAI Astraのリリースが間近に迫る中、研究者たちはそのアーキテクチャの監視がより困難になることを懸念しています","summary":"OpenAIは安全プロトコルを強化するために最強モデルであるAstraのリリースを遅らせました。The Informationによると、Astraはより不透明なリカレント深度/ループトランス技術を採用し、内部思考の監視が困難になる可能性があるとのことです。","category":"業界動向","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"OpenAI Astraのリリースが間近に迫る中、研究者たちはそのアーキテクチャの監視がより困難になることを懸念しています - Aioga AIニュース","description":"OpenAIは安全プロトコルを強化するために最強モデルであるAstraのリリースを遅らせました。The Informationによると、Astraはより不透明なリカレント深度/ループトランス技術を採用し、内部思考の監視が困難になる可能性があるとのことです。","url":"https://www.aioga.com/ja/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:13.869Z"},"ko":{"title":"OpenAI Astra의 출시가 임박한 가운데, 연구진은 그 아키텍처가 감시하기 더 어려워질 것이라 우려하고 있습니다","summary":"OpenAI는 안전 프로토콜을 강화하기 위해 가장 강력한 모델인 Astra 출시를 연기했습니다. The Information은 Astra가 더 불투명한 반복 깊이/루프 변압기 기술을 채택할 수 있어 내부 사고 모니터링이 어려워질 수 있다고 보도했습니다.","category":"업계 동향","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"OpenAI Astra의 출시가 임박한 가운데, 연구진은 그 아키텍처가 감시하기 더 어려워질 것이라 우려하고 있습니다 - Aioga AI 뉴스","description":"OpenAI는 안전 프로토콜을 강화하기 위해 가장 강력한 모델인 Astra 출시를 연기했습니다. The Information은 Astra가 더 불투명한 반복 깊이/루프 변압기 기술을 채택할 수 있어 내부 사고 모니터링이 어려워질 수 있다고 보도했습니다.","url":"https://www.aioga.com/ko/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:22.384Z"},"es":{"title":"Con el lanzamiento inminente de OpenAI Astra, los investigadores temen que su arquitectura sea más difícil de monitorizar","summary":"OpenAI retrasó el lanzamiento de su modelo más potente, Astra, para reforzar los protocolos de seguridad. The Information informó que Astra podría adoptar una tecnología de transformador de profundidad/bucle recurrente más opaca, dificultando la monitorización del pensamiento interno.","category":"Industria","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Con el lanzamiento inminente de OpenAI Astra, los investigadores temen que su arquitectura sea más difícil de monitorizar - Aioga Noticias de IA","description":"OpenAI retrasó el lanzamiento de su modelo más potente, Astra, para reforzar los protocolos de seguridad. The Information informó que Astra podría adoptar una tecnología de transfo...","url":"https://www.aioga.com/es/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:22.833Z"},"fr":{"title":"Avec la sortie imminente d’OpenAI Astra, les chercheurs craignent que son architecture ne soit plus difficile à surveiller","summary":"OpenAI a retardé la sortie de son modèle le plus performant, Astra, afin de renforcer les protocoles de sécurité. The Information a rapporté qu’Astra pourrait adopter une technologie de transformateur à profondeur récurrente/boucle plus opaque, rendant la réflexion interne plus difficile à surveiller.","category":"Industrie","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Avec la sortie imminente d’OpenAI Astra, les chercheurs craignent que son architecture ne soit plus difficile à surveiller - Aioga Actualités IA","description":"OpenAI a retardé la sortie de son modèle le plus performant, Astra, afin de renforcer les protocoles de sécurité. The Information a rapporté qu’Astra pourrait adopter une technolog...","url":"https://www.aioga.com/fr/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:31.890Z"},"de":{"title":"Mit der bevorstehenden Veröffentlichung von OpenAI Astra befürchten Forscher, dass seine Architektur schwerer zu überwachen sein könnte","summary":"OpenAI hat die Veröffentlichung seines stärksten Modells, Astra, verzögert, um Sicherheitsprotokolle zu stärken. The Information berichtete, dass Astra möglicherweise eine mehr undurchsichtige Repeater-Tiefen-/Schleifen-Transformator-Technologie übernehmen könnte, was das interne Überwachen erschwert.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Mit der bevorstehenden Veröffentlichung von OpenAI Astra befürchten Forscher, dass seine Architektur schwerer zu überwachen sein könnte - Aioga KI-News","description":"OpenAI hat die Veröffentlichung seines stärksten Modells, Astra, verzögert, um Sicherheitsprotokolle zu stärken. The Information berichtete, dass Astra möglicherweise eine mehr und...","url":"https://www.aioga.com/de/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:31.928Z"},"pt-BR":{"title":"Com o lançamento iminente do OpenAI Astra, pesquisadores estão preocupados que sua arquitetura seja mais difícil de monitorar","summary":"A OpenAI adiou o lançamento de seu modelo mais forte, o Astra, para fortalecer os protocolos de segurança. A Informação informou que a Astra pode adotar uma tecnologia de transformador de profundidade/loop recorrente mais opaca, tornando o pensamento interno mais difícil de monitorar.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Com o lançamento iminente do OpenAI Astra, pesquisadores estão preocupados que sua arquitetura seja mais difícil de monitorar - Aioga Notícias de IA","description":"A OpenAI adiou o lançamento de seu modelo mais forte, o Astra, para fortalecer os protocolos de segurança. A Informação informou que a Astra pode adotar uma tecnologia de transform...","url":"https://www.aioga.com/pt-BR/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:41.043Z"},"ru":{"title":"С предстоящим выходом OpenAI Astra исследователи обеспокоены тем, что её архитектуру будет сложнее контролировать","summary":"OpenAI отложила выпуск своей самой сильной модели — Astra, чтобы усилить протоколы безопасности. The Information сообщила, что Astra может использовать более непрозрачную технологию рекуррентной глубины/петля трансформатора, что усложнит мониторинг внутреннего мышления.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"С предстоящим выходом OpenAI Astra исследователи обеспокоены тем, что её архитектуру будет сложнее контролировать - Aioga Новости ИИ","description":"OpenAI отложила выпуск своей самой сильной модели — Astra, чтобы усилить протоколы безопасности. The Information сообщила, что Astra может использовать более непрозрачную технологи...","url":"https://www.aioga.com/ru/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:39.521Z"},"ar":{"title":"مع اقتراب إصدار OpenAI Astra، يشعر الباحثون بالقلق من أن بنيته ستكون أصعب في المراقبة","summary":"أخرت OpenAI إصدار أقوى نموذج لها، أسترا، لتعزيز بروتوكولات السلامة. ذكرت صحيفة ذا إنفورميشن أن أسترا قد تعتمد تقنية عمق متكرر/محول حلقي أكثر غموضا، مما يجعل من الصعب مراقبة التفكير الداخلي.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"مع اقتراب إصدار OpenAI Astra، يشعر الباحثون بالقلق من أن بنيته ستكون أصعب في المراقبة - Aioga أخبار الذكاء الاصطناعي","description":"أخرت OpenAI إصدار أقوى نموذج لها، أسترا، لتعزيز بروتوكولات السلامة. ذكرت صحيفة ذا إنفورميشن أن أسترا قد تعتمد تقنية عمق متكرر/محول حلقي أكثر غموضا، مما يجعل من الصعب مراقبة التفكير...","url":"https://www.aioga.com/ar/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:49.985Z"},"hi":{"title":"OpenAI एस्ट्रा की रिलीज़ के साथ, शोधकर्ता चिंतित हैं कि इसकी वास्तुकला की निगरानी करना कठिन होगा","summary":"OpenAI ने सुरक्षा प्रोटोकॉल को मजबूत करने के लिए अपने सबसे मजबूत मॉडल, एस्ट्रा को जारी करने में देरी की। जानकारी ने बताया कि एस्ट्रा अधिक अपारदर्शी आवर्तक गहराई/लूप ट्रांसफार्मर तकनीक को अपना सकता है, जिससे आंतरिक सोच की निगरानी करना कठिन हो जाएगा।","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"OpenAI एस्ट्रा की रिलीज़ के साथ, शोधकर्ता चिंतित हैं कि इसकी वास्तुकला की निगरानी करना कठिन होगा - Aioga AI समाचार","description":"OpenAI ने सुरक्षा प्रोटोकॉल को मजबूत करने के लिए अपने सबसे मजबूत मॉडल, एस्ट्रा को जारी करने में देरी की। जानकारी ने बताया कि एस्ट्रा अधिक अपारदर्शी आवर्तक गहराई/लूप ट्रांसफार्मर तक...","url":"https://www.aioga.com/hi/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:49.959Z"},"it":{"title":"Con l'imminente rilascio di OpenAI Astra, i ricercatori temono che la sua architettura sarà più difficile da monitorare","summary":"OpenAI ha ritardato il rilascio del suo modello più potente, Astra, per rafforzare i protocolli di sicurezza. The Information ha riportato che Astra potrebbe adottare una tecnologia di trasformatore a profondità/loop ricorrente più opaca, rendendo il pensiero interno più difficile da monitorare.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Con l'imminente rilascio di OpenAI Astra, i ricercatori temono che la sua architettura sarà più difficile da monitorare - Aioga Notizie IA","description":"OpenAI ha ritardato il rilascio del suo modello più potente, Astra, per rafforzare i protocolli di sicurezza. The Information ha riportato che Astra potrebbe adottare una tecnologi...","url":"https://www.aioga.com/it/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:58.974Z"},"nl":{"title":"Met de naderende release van OpenAI Astra maken onderzoekers zich zorgen dat de architectuur moeilijker te monitoren zal zijn","summary":"OpenAI stelde de uitgave van zijn sterkste model, Astra, uit om de veiligheidsprotocollen te versterken. The Information meldde dat Astra mogelijk een meer ondoorzichtige transformator met terugkerende diepte/looping zal gebruiken, waardoor intern denken moeilijker te monitoren is.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Met de naderende release van OpenAI Astra maken onderzoekers zich zorgen dat de architectuur moeilijker te monitoren zal zijn - Aioga AI-nieuws","description":"OpenAI stelde de uitgave van zijn sterkste model, Astra, uit om de veiligheidsprotocollen te versterken. The Information meldde dat Astra mogelijk een meer ondoorzichtige transform...","url":"https://www.aioga.com/nl/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:44:59.080Z"},"tr":{"title":"OpenAI Astra'nın yakında çıkışı yaklaşırken, araştırmacılar mimarisinin izlenmesinin daha zor olacağından endişe ediyorlar","summary":"OpenAI, güvenlik protokollerini güçlendirmek için en güçlü modeli Astra'yı yayınlamayı geciktirdi. The Information, Astra'nın daha opak tekrarlayan derinlik/döngülü transformatör teknolojisini benimseyebileceğini ve bunun da iç düşünceyi izlemeyi zorlaştırabileceğini bildirdi.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"OpenAI Astra'nın yakında çıkışı yaklaşırken, araştırmacılar mimarisinin izlenmesinin daha zor olacağından endişe ediyorlar - Aioga AI Haberleri","description":"OpenAI, güvenlik protokollerini güçlendirmek için en güçlü modeli Astra'yı yayınlamayı geciktirdi. The Information, Astra'nın daha opak tekrarlayan derinlik/döngülü transformatör t...","url":"https://www.aioga.com/tr/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:45:07.692Z"},"vi":{"title":"Với việc OpenAI Astra sắp ra mắt, các nhà nghiên cứu lo ngại rằng kiến trúc của nó sẽ khó giám sát hơn","summary":"OpenAI đã trì hoãn việc phát hành mô hình mạnh nhất của mình, Astra, để tăng cường các giao thức an toàn. The Information cho biết Astra có thể áp dụng công nghệ biến áp chiều sâu lặp lại/lặp lại mơ hồ hơn, khiến việc theo dõi tư duy nội bộ trở nên khó khăn hơn.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Với việc OpenAI Astra sắp ra mắt, các nhà nghiên cứu lo ngại rằng kiến trúc của nó sẽ khó giám sát hơn - Tin tức AI Aioga","description":"OpenAI đã trì hoãn việc phát hành mô hình mạnh nhất của mình, Astra, để tăng cường các giao thức an toàn. The Information cho biết Astra có thể áp dụng công nghệ biến áp chiều sâu...","url":"https://www.aioga.com/vi/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:45:07.637Z"},"id":{"title":"Dengan peluncuran OpenAI Astra yang sudah dekat, para peneliti khawatir arsitekturnya akan lebih sulit dipantau","summary":"OpenAI menunda peluncuran model terkuatnya, Astra, untuk memperkuat protokol keselamatan. The Information melaporkan bahwa Astra mungkin akan mengadopsi teknologi trafo kedalaman berulang/loop yang lebih buram, sehingga pemikiran internal menjadi lebih sulit dipantau.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"Dengan peluncuran OpenAI Astra yang sudah dekat, para peneliti khawatir arsitekturnya akan lebih sulit dipantau - Berita AI Aioga","description":"OpenAI menunda peluncuran model terkuatnya, Astra, untuk memperkuat protokol keselamatan. The Information melaporkan bahwa Astra mungkin akan mengadopsi teknologi trafo kedalaman b...","url":"https://www.aioga.com/id/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:45:16.015Z"},"th":{"title":"เมื่อ OpenAI Astra กําลังจะเปิดตัว นักวิจัยกังวลว่าสถาปัตยกรรมของมันจะยากต่อการตรวจสอบมากขึ้น","summary":"OpenAI เลื่อนการเปิดตัวโมเดลที่แข็งแกร่งที่สุดคือ Astra เพื่อเสริมสร้างมาตรการความปลอดภัย Information รายงานว่า Astra อาจนําเทคโนโลยีหม้อแปลงแบบลึก/ลูปที่ทึบแสงมากขึ้น ทําให้การคิดภายในยากต่อการตรวจสอบ","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"เมื่อ OpenAI Astra กําลังจะเปิดตัว นักวิจัยกังวลว่าสถาปัตยกรรมของมันจะยากต่อการตรวจสอบมากขึ้น - ข่าว AI Aioga","description":"OpenAI เลื่อนการเปิดตัวโมเดลที่แข็งแกร่งที่สุดคือ Astra เพื่อเสริมสร้างมาตรการความปลอดภัย Information รายงานว่า Astra อาจนําเทคโนโลยีหม้อแปลงแบบลึก/ลูปที่ทึบแสงมากขึ้น ทําให้การคิด...","url":"https://www.aioga.com/th/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:45:15.914Z"},"pl":{"title":"W związku z nadchodzącą premierą OpenAI Astra naukowcy obawiają się, że jej architektura będzie trudniejsza do monitorowania","summary":"OpenAI opóźniło wydanie swojego najsilniejszego modelu, Astra, aby wzmocnić protokoły bezpieczeństwa. Informacja informowała, że Astra może przyjąć bardziej nieprzejrzystą technologię transformatorów z głębokością rekurencyjną/pętlami, co utrudnia monitorowanie myślenia wewnętrznego.","category":"行业动态","source":"The Verge：AI（RSS）","aggregationSource":"The Verge：AI（RSS）","pageTitle":"W związku z nadchodzącą premierą OpenAI Astra naukowcy obawiają się, że jej architektura będzie trudniejsza do monitorowania - Aioga Wiadomości AI","description":"OpenAI opóźniło wydanie swojego najsilniejszego modelu, Astra, aby wzmocnić protokoły bezpieczeństwa. Informacja informowała, że Astra może przyjąć bardziej nieprzejrzystą technolo...","url":"https://www.aioga.com/pl/news/cmtkcfd1q018grog03eluywgr/","contentTranslated":true,"sourceHash":"7a70f848149198b4","translatedAt":"2026-09-02T17:45:25.086Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}