{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-09-20T15:01:03.420Z","headline":"OpenAI 新推理技术引发 AI 安全专家担忧：Astra 采用不透明循环或影响链式思维监控","description":"据 The Information 报道，OpenAI 新模型 Astra 将使用名为 recurrent depth（不透明循环）的推理技术，让模型以循环方式多次处理同一查询，留下更少的可读痕迹，可能使链式思维更难监控。","url":"https://www.aioga.com/news/cmtkjxgun03wqroll8e8qykqg/","mainEntityOfPage":"https://www.aioga.com/news/cmtkjxgun03wqroll8e8qykqg/","datePublished":"2026-09-02T20:19:14.000Z","dateModified":"2026-09-02T20:19:14.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts","https://aihot.virxact.com/items/cmtkjxgun03wqroll8e8qykqg"],"canonicalUrl":"https://www.aioga.com/news/cmtkjxgun03wqroll8e8qykqg/","directAnswer":{"@type":"Answer","text":"据报道，OpenAI 新模型 Astra 将使用 recurrent depth，也称 opaque recurrence，使模型以循环方式多次处理同一查询，并留下较少可读痕迹。报道认为，这可能使链式思维更难监控；该技术在 Astra 中的使用似乎有限。","url":"https://www.aioga.com/news/cmtkjxgun03wqroll8e8qykqg/","dateCreated":"2026-09-02T20:19:14.000Z","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":"TechCrunch source article","url":"https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts","datePublished":"2026-09-02T20:19:14.000Z","provider":{"@type":"Organization","name":"TechCrunch","url":"https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmtkjxgun03wqroll8e8qykqg","datePublished":"2026-09-02T20:19:14.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmtkjxgun03wqroll8e8qykqg"}}],"aggregationSource":"TechCrunch：AI（RSS）","originalPublisher":{"name":"TechCrunch","url":"https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts"},"geoDeepAnswer":null,"article":{"id":"cmtkjxgun03wqroll8e8qykqg","slug":"cmtkjxgun03wqroll8e8qykqg","url":"https://www.aioga.com/news/cmtkjxgun03wqroll8e8qykqg/","title":"OpenAI 新推理技术引发 AI 安全专家担忧：Astra 采用不透明循环或影响链式思维监控","title_en":"","summary":"据 The Information 报道，OpenAI 新模型 Astra 将使用名为 recurrent depth（不透明循环）的推理技术，让模型以循环方式多次处理同一查询，留下更少的可读痕迹，可能使链式思维更难监控。","source":"TechCrunch：AI（RSS）","sourceUrl":"https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts","aiHotUrl":"https://aihot.virxact.com/items/cmtkjxgun03wqroll8e8qykqg","publishedAt":"2026-09-02T20:19:14.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["OpenAI’s new Astra model will use a reasoning technique called “recurrent depth” that allows it to operate outside of the sequential thinking that characterizes most reasoning models, the The Information reported：https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns on Tuesday. This technique, also called “opaque recurrence,” will likely make the model’s chain of thought more difficult to monitor — and that has AI safety experts rattled.","While Astra’s use of the technique is reportedly limited, its emergence has still raised significant concerns among AI safety experts.","“I am extremely concerned by the reporting that Astra uses opaque recurrence,” wrote Redwood CEO Buck Shlegeris in a post：https://x.com/bshlgrs/status/2094990313513439464?s=46&t=45_xAnRsdQP1GVqYv9Gdbw after the news broke. “I don’t know whether Astra is much less CoT monitorable than previous models. But if OpenAI pushes this technique further, they’ll have the option to massively increase the recurrence and totally destroys CoT monitorability.”","Longtime AI safety advocate Zvi Mowshowitz also weighed in and wrote：https://thezvi.substack.com/p/anthropic-has-some-alignment-problems?open=false#%C2%A7this-just-in that laws might be necessary to prevent a “race to the bottom” among AI labs.","“The technique is playing with fire, risking a taboo that OpenAI and Anthropic have fought to establish that we work hard to maintain Chain of Thought faithfulness and monitorability for as long as we can,” Mowshowitz wrote. “More intensive use of such techniques would probably damage monitorability.”","Under normal circumstances, a reasoning model’s chain of thought provides the sequential steps taken by the model as it attempts to solve a problem. While the representation is imperfect, it still serves as a valuable tool for monitoring misbehavior or misalignment. In the case of OpenAI’s recent rogue agent activity, chain-of-thought records were an important tool in teasing out why agents behaved the way they did.","In opaque recurrence, the model takes a less linear approach, processing the same query several times in a loop. The result leaves fewer legible traces, effectively side-stepping a conventional chain-of-thought record.","Crucially, Astra’s use of the technique appears to be limited. The model’s chain of thought is still expected to be legible, and the company pushed back against any suggestion that it would shift to “neuralese.” OpenAI has already announced plans for extensive chain-of-thought monitoring systems as part of its forward-looking safety plans.","In a post on X：https://x.com/merettm/status/2095023204993490967, OpenAI chief scientist Jakub Pachocki emphasized the lab’s commitment to legible chains of thought. “OpenAI has worked to preserve and utilize chain-of-thought monitoring since our very first reasoning models,” Pachocki wrote. “It’s a core goal of our current research program.","All AI models do some quantity of opaque reasoning, and few researchers take chain-of-thought logs as a direct representation of a model’s reasoning. Still, those caveats don’t dispel the concern that opaque recurrence may make AI reasoning harder to monitor, particularly as it grows in use across different models. In a follow-up report Wednesday morning, The Information reported that both Anthropic and Google DeepMind were already discussing the technique.","In a post responding to the news,：https://x.com/ryangreenblatt/status/2094996656186081642?s=46&t=45_xAnRsdQP1GVqYv9Gdbw Redwood Research chief scientist Ryan Greenblatt said opaque reasoning could easily scale faster than conventional chain-of-thought reasoning, effectively removing all reasoning from visible channels.","“My biggest concern is that a natural progression from here would involve scaling up the opaque reasoning to the point where the model reasons entirely or almost entirely in latent space,” Greenblatt wrote. “I hope it isn’t too late to avoid the most concerning architectures and that OpenAI will stop here.”","When you purchase through links in our articles, we may earn a small commission：https://techcrunch.com/techcrunch-affiliate-monetization-standards/. This doesn’t affect our editorial independence.","Don’t miss out . The startup community will gather to answer a pivotal question: How do you build sustainably in the AI era?","Uber is laying off 10% of staff, or 3,300 people：https://techcrunch.com/2026/09/02/uber-is-laying-off-10-of-staff-or-3300-people/ Ram Iyer：https://techcrunch.com/author/ram-iyer/","Apple shares ‘shocking evidence’ against former employee accused of stealing company data for OpenAI：https://techcrunch.com/2026/08/31/apple-shares-shocking-evidence-against-former-employee-accused-of-stealing-company-data-for-openai/ Amanda Silberling：https://techcrunch.com/author/amanda-silberling/","Microsoft tests fix for latest hours-long Outlook outage：https://techcrunch.com/2026/08/31/microsoft-tests-fix-for-latest-hours-long-outlook-outage/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","MapQuest’s app surges to No. 1 in Navigation after refusing to rename Lake Ontario：https://techcrunch.com/2026/08/31/mapquests-app-surges-to-no-1-in-navigation-after-refusing-to-rename-lake-ontario/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","Musk’s faster path to more gas turbines comes with pollution problem：https://techcrunch.com/2026/08/30/musks-faster-path-to-more-gas-turbines-comes-with-pollution-problem/ Connie Loizos：https://techcrunch.com/author/connie-loizos/","Hugging Face is selling a cute $399 open source duck robot, Microduck：https://techcrunch.com/2026/08/27/hugging-face-is-selling-a-cute-399-open-source-duck-robot-microduck/ Rebecca Bellan：https://techcrunch.com/author/rebecca-bellan/","Nvidia closes in on Hugging Face acquisition：https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/ Connie Loizos：https://techcrunch.com/author/connie-loizos/"],"articleImages":[{"sourceUrl":"https://techcrunch.com/wp-content/uploads/2025/05/russell-e1755718978143.jpg?w=150","alt":"Russell Brandom","afterParagraph":12,"url":"/media/articles/cmtkjxgun03wqroll8e8qykqg/3ca54d84294be5e7.webp"},{"sourceUrl":"https://techcrunch.com/wp-content/uploads/2025/04/Disrupt2026-Color.png","alt":"Event Logo","afterParagraph":12,"url":"/media/articles/cmtkjxgun03wqroll8e8qykqg/d1d06f6f97361f52.webp"}],"mediaStatus":"ok","articleBodyZh":["OpenAI的新Astra模型将使用一种称为“递归深度”的推理技术，使其能够超越大多数推理模型所特有的顺序思维，《The Information》周二报道：https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns。这种技术，也被称为“不透明递归”，可能会使模型的思维链更难以监控——这让人工智能安全专家感到不安。","尽管据报道Astra对该技术的使用有限，但其出现仍引起了人工智能安全专家的重大关注。","“我对于Astra使用不透明递归的报道感到极度担忧，”Redwood首席执行官Buck Shlegeris在消息发布后写道：https://x.com/bshlgrs/status/2094990313513439464?s=46&t=45_xAnRsdQP1GVqYv9Gdbw。“我不知道Astra是否比以往的模型更难以监控思维链。但如果OpenAI进一步推进这一技术，他们将有机会大幅增加递归，从而完全破坏思维链的可监控性。”","长期从事人工智能安全倡导的Zvi Mowshowitz也发表了看法，并写道：https://thezvi.substack.com/p/anthropic-has-some-alignment-problems?open=false#%C2%A7this-just-in 可能有必要制定法律，以防止AI实验室之间的“恶性竞争”。","“Mowshowitz写道，这项技术是在玩火，冒着触碰禁忌的风险——OpenAI和Anthropic一直在努力建立并维持思维链的忠实性和可监控性。” Mowshowitz写道。“更频繁地使用此类技术可能会损害可监控性。”","在正常情况下，推理模型的思维链提供了模型试图解决问题时采取的顺序步骤。虽然这种表述并不完美，但它仍然是监控模型行为异常或不匹配的重要工具。在OpenAI近期的流氓代理活动中，思维链记录是揭示代理行为原因的重要工具。","在不透明递归中，模型采用较不线性的方式，多次循环处理同一查询。结果留下的可读痕迹较少，有效地绕过了传统的思维链记录。","关键是，Astra 对该技术的使用似乎是有限的。模型的思维链仍然应当是可读的，公司也对任何将其转向“神经语” 的建议表示反对。OpenAI 已经宣布计划建立广泛的链式思维监控系统，作为其前瞻性安全计划的一部分。","在 X 上的一篇帖子中：https://x.com/merettm/status/2095023204993490967，OpenAI 首席科学家 Jakub Pachocki 强调了实验室对可读思维链的承诺。Pachocki 写道：“自从我们最早的推理模型开始，OpenAI 就一直致力于保存和利用思维链监控。这是我们当前研究计划的核心目标。”","所有 AI 模型都进行一定程度的不透明推理，很少有研究人员将思维链日志视为模型推理的直接表现。然而，这些提示并不能消除人们对不透明递归可能使 AI 推理更难监控的担忧，尤其当它在不同模型中越来越广泛使用时。在周三早上的后续报道中，《The Information》报道称，Anthropic 和 Google DeepMind 已经在讨论该技术。","在回应该新闻的一篇帖子中：https://x.com/ryangreenblatt/status/2094996656186081642?s=46&t=45_xAnRsdQP1GVqYv9Gdbw，Redwood Research 首席科学家 Ryan Greenblatt 表示，不透明推理可能比传统链式思维推理增长得更快，有效地将所有推理从可见渠道中移除。","Greenblatt 写道：“我最担心的是，从这里自然发展的路径可能会导致不透明推理扩展到模型完全或几乎完全在潜在空间中推理的程度。我希望避免最令人担忧的架构还不算太晚，也希望 OpenAI 会在此止步。”","当您通过我们文章中的链接购买时，我们可能会赚取少量佣金：https://techcrunch.com/techcrunch-affiliate-monetization-standards/。这不会影响我们的编辑独立性。","不要错过。创业社区将齐聚一堂，回答一个关键问题：在 AI 时代如何实现可持续建设？","优步正在裁员10%，即3300人：https://techcrunch.com/2026/09/02/uber-is-laying-off-10-of-staff-or-3300-people/ Ram Iyer：https://techcrunch.com/author/ram-iyer/","苹果分享针对前员工的“令人震惊的证据”，该员工被指控为OpenAI窃取公司数据：https://techcrunch.com/2026/08/31/apple-shares-shocking-evidence-against-former-employee-accused-of-stealing-company-data-for-openai/ Amanda Silberling：https://techcrunch.com/author/amanda-silberling/","微软测试解决最新长时间Outlook宕机的方法：https://techcrunch.com/2026/08/31/microsoft-tests-fix-for-latest-hours-long-outlook-outage/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","MapQuest应用在拒绝重命名安大略湖后跃升至导航类第一：https://techcrunch.com/2026/08/31/mapquests-app-surges-to-no-1-in-navigation-after-refusing-to-rename-lake-ontario/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","马斯克加快更多燃气轮机进程但带来污染问题：https://techcrunch.com/2026/08/30/musks-faster-path-to-more-gas-turbines-comes-with-pollution-problem/ Connie Loizos：https://techcrunch.com/author/connie-loizos/","Hugging Face正在销售一款可爱的399美元开源小鸭机器人，Microduck：https://techcrunch.com/2026/08/27/hugging-face-is-selling-a-cute-399-open-source-duck-robot-microduck/ Rebecca Bellan：https://techcrunch.com/author/rebecca-bellan/","Nvidia 接近收购 Hugging Face：https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/ Connie Loizos：https://techcrunch.com/author/connie-loizos/"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"据报道，OpenAI 新模型 Astra 将使用 recurrent depth，也称 opaque recurrence，使模型以循环方式多次处理同一查询，并留下较少可读痕迹。报道认为，这可能使链式思维更难监控；该技术在 Astra 中的使用似乎有限。","background":"通常情况下，推理模型的链式思维会呈现解决问题时的连续步骤，虽不完美，仍可作为监测模型行为的工具。opaque recurrence 采用较少线性的循环处理方式，可能减少传统链式思维记录中的可读痕迹。","viewpoint":"Aioga 判断：来源呈现的是对监控难度的担忧，而非 Astra 已无法监控的事实。专家关注该技术若被进一步强化，可能削弱链式思维的可监控性；但材料同时称其当前使用似乎有限。","implications":"可能影响：若该技术被更密集地使用，链式思维监控可能面临更大挑战，需要结合更多安全监测机制。Astra 当前使用似乎有限，不代表其链式思维已经不可读，也不足以据此断定实际安全影响。","nextStep":"后续观察：应关注 OpenAI 对 Astra 使用范围、链式思维可读性及监控系统的后续说明，也需核查后续材料是否提供关于该技术实际安全影响的证据。","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":"2026-09-06T05:43:46.696Z","sourceHash":"aa4558c32aa5c7bb","review":{"approved":true,"groundedness":95,"clarity":92,"duplicationRisk":10,"blockingIssues":[],"notes":["“Aioga 判断”属于编辑框架性表述，未构成事实错误；如需更客观，可改为“材料显示”。","“需要结合更多安全监测机制”属于合理的推论或建议，来源未直接提出，但已使用“可能影响”语境，未冒充已证实事实。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":1,"checks":["schema","length","source-attribution","editorial-labels","inference-boundary","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","TechCrunch：AI（RSS）"],"translations":{"zh-CN":{"title":"OpenAI 新推理技术引发 AI 安全专家担忧：Astra 采用不透明循环或影响链式思维监控","summary":"据 The Information 报道，OpenAI 新模型 Astra 将使用名为 recurrent depth（不透明循环）的推理技术，让模型以循环方式多次处理同一查询，留下更少的可读痕迹，可能使链式思维更难监控。","category":"行业动态","source":"TechCrunch","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI 新推理技术引发 AI 安全专家担忧：Astra 采用不透明循环或影响链式思维监控 - Aioga AI资讯","description":"据 The Information 报道，OpenAI 新模型 Astra 将使用名为 recurrent depth（不透明循环）的推理技术，让模型以循环方式多次处理同一查询，留下更少的可读痕迹，可能使链式思维更难监控。","url":"https://www.aioga.com/news/cmtkjxgun03wqroll8e8qykqg/","articleBody":["OpenAI的新Astra模型将使用一种称为“递归深度”的推理技术，使其能够超越大多数推理模型所特有的顺序思维，《The Information》周二报道：https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns。这种技术，也被称为“不透明递归”，可能会使模型的思维链更难以监控——这让人工智能安全专家感到不安。","尽管据报道Astra对该技术的使用有限，但其出现仍引起了人工智能安全专家的重大关注。","“我对于Astra使用不透明递归的报道感到极度担忧，”Redwood首席执行官Buck Shlegeris在消息发布后写道：https://x.com/bshlgrs/status/2094990313513439464?s=46&t=45_xAnRsdQP1GVqYv9Gdbw。“我不知道Astra是否比以往的模型更难以监控思维链。但如果OpenAI进一步推进这一技术，他们将有机会大幅增加递归，从而完全破坏思维链的可监控性。”","长期从事人工智能安全倡导的Zvi Mowshowitz也发表了看法，并写道：https://thezvi.substack.com/p/anthropic-has-some-alignment-problems?open=false#%C2%A7this-just-in 可能有必要制定法律，以防止AI实验室之间的“恶性竞争”。","“Mowshowitz写道，这项技术是在玩火，冒着触碰禁忌的风险——OpenAI和Anthropic一直在努力建立并维持思维链的忠实性和可监控性。” Mowshowitz写道。“更频繁地使用此类技术可能会损害可监控性。”","在正常情况下，推理模型的思维链提供了模型试图解决问题时采取的顺序步骤。虽然这种表述并不完美，但它仍然是监控模型行为异常或不匹配的重要工具。在OpenAI近期的流氓代理活动中，思维链记录是揭示代理行为原因的重要工具。","在不透明递归中，模型采用较不线性的方式，多次循环处理同一查询。结果留下的可读痕迹较少，有效地绕过了传统的思维链记录。","关键是，Astra 对该技术的使用似乎是有限的。模型的思维链仍然应当是可读的，公司也对任何将其转向“神经语” 的建议表示反对。OpenAI 已经宣布计划建立广泛的链式思维监控系统，作为其前瞻性安全计划的一部分。","在 X 上的一篇帖子中：https://x.com/merettm/status/2095023204993490967，OpenAI 首席科学家 Jakub Pachocki 强调了实验室对可读思维链的承诺。Pachocki 写道：“自从我们最早的推理模型开始，OpenAI 就一直致力于保存和利用思维链监控。这是我们当前研究计划的核心目标。”","所有 AI 模型都进行一定程度的不透明推理，很少有研究人员将思维链日志视为模型推理的直接表现。然而，这些提示并不能消除人们对不透明递归可能使 AI 推理更难监控的担忧，尤其当它在不同模型中越来越广泛使用时。在周三早上的后续报道中，《The Information》报道称，Anthropic 和 Google DeepMind 已经在讨论该技术。","在回应该新闻的一篇帖子中：https://x.com/ryangreenblatt/status/2094996656186081642?s=46&t=45_xAnRsdQP1GVqYv9Gdbw，Redwood Research 首席科学家 Ryan Greenblatt 表示，不透明推理可能比传统链式思维推理增长得更快，有效地将所有推理从可见渠道中移除。","Greenblatt 写道：“我最担心的是，从这里自然发展的路径可能会导致不透明推理扩展到模型完全或几乎完全在潜在空间中推理的程度。我希望避免最令人担忧的架构还不算太晚，也希望 OpenAI 会在此止步。”","当您通过我们文章中的链接购买时，我们可能会赚取少量佣金：https://techcrunch.com/techcrunch-affiliate-monetization-standards/。这不会影响我们的编辑独立性。","不要错过。创业社区将齐聚一堂，回答一个关键问题：在 AI 时代如何实现可持续建设？","优步正在裁员10%，即3300人：https://techcrunch.com/2026/09/02/uber-is-laying-off-10-of-staff-or-3300-people/ Ram Iyer：https://techcrunch.com/author/ram-iyer/","苹果分享针对前员工的“令人震惊的证据”，该员工被指控为OpenAI窃取公司数据：https://techcrunch.com/2026/08/31/apple-shares-shocking-evidence-against-former-employee-accused-of-stealing-company-data-for-openai/ Amanda Silberling：https://techcrunch.com/author/amanda-silberling/","微软测试解决最新长时间Outlook宕机的方法：https://techcrunch.com/2026/08/31/microsoft-tests-fix-for-latest-hours-long-outlook-outage/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","MapQuest应用在拒绝重命名安大略湖后跃升至导航类第一：https://techcrunch.com/2026/08/31/mapquests-app-surges-to-no-1-in-navigation-after-refusing-to-rename-lake-ontario/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","马斯克加快更多燃气轮机进程但带来污染问题：https://techcrunch.com/2026/08/30/musks-faster-path-to-more-gas-turbines-comes-with-pollution-problem/ Connie Loizos：https://techcrunch.com/author/connie-loizos/","Hugging Face正在销售一款可爱的399美元开源小鸭机器人，Microduck：https://techcrunch.com/2026/08/27/hugging-face-is-selling-a-cute-399-open-source-duck-robot-microduck/ Rebecca Bellan：https://techcrunch.com/author/rebecca-bellan/","Nvidia 接近收购 Hugging Face：https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/ Connie Loizos：https://techcrunch.com/author/connie-loizos/"]},"en":{"title":"OpenAI's new reasoning technology raises concerns among AI safety experts: Astra's use of opaque loops may affect chain-of-thought monitoring","summary":"According to The Information, OpenAI's new model Astra will use a reasoning technique called recurrent depth, allowing the model to process the same query multiple times in a loop, leaving fewer readable traces, which may make chain-of-thought reasoning harder to monitor.","category":"Industry","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI's new reasoning technology raises concerns among AI safety experts: Astra's use of opaque loops may affect chain-of-thought monitoring - Aioga AI News","description":"According to The Information, OpenAI's new model Astra will use a reasoning technique called recurrent depth, allowing the model to process the same query multiple times in a loop,...","url":"https://www.aioga.com/en/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:42:22.432Z"},"ja":{"title":"OpenAI の新しい推論技術が AI セキュリティ専門家の懸念を引き起こす：Astra は不透明なループを採用しており、連鎖的思考の監視に影響を与える可能性がある","summary":"The Informationの報道によると、OpenAIの新モデルAstraは、recurrent depth（不透明な循環）と呼ばれる推論技術を使用し、モデルが同じクエリを循環的に何度も処理して、読み取れる痕跡を残さず、連鎖思考の監視をより困難にする可能性がある。","category":"業界動向","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI の新しい推論技術が AI セキュリティ専門家の懸念を引き起こす：Astra は不透明なループを採用しており、連鎖的思考の監視に影響を与える可能性がある - Aioga AIニュース","description":"The Informationの報道によると、OpenAIの新モデルAstraは、recurrent depth（不透明な循環）と呼ばれる推論技術を使用し、モデルが同じクエリを循環的に何度も処理して、読み取れる痕跡を残さず、連鎖思考の監視をより困難にする可能性がある。","url":"https://www.aioga.com/ja/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:42:32.277Z"},"ko":{"title":"OpenAI의 새로운 추론 기술, AI 안전 전문가들의 우려 불러: Astra의 불투명한 루프 사용이 연쇄적 사고 모니터링에 영향 줄 수 있어","summary":"The Information에 따르면, OpenAI의 새 모델 Astra는 recurrent depth(불투명 순환)이라는 추론 기술을 사용하여 모델이 같은 쿼리를 여러 번 순환 방식으로 처리하게 하며, 읽을 수 있는 흔적을 덜 남겨 체인 사고를 더 어렵게 모니터링할 수 있다고 합니다.","category":"업계 동향","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI의 새로운 추론 기술, AI 안전 전문가들의 우려 불러: Astra의 불투명한 루프 사용이 연쇄적 사고 모니터링에 영향 줄 수 있어 - Aioga AI 뉴스","description":"The Information에 따르면, OpenAI의 새 모델 Astra는 recurrent depth(불투명 순환)이라는 추론 기술을 사용하여 모델이 같은 쿼리를 여러 번 순환 방식으로 처리하게 하며, 읽을 수 있는 흔적을 덜 남겨 체인 사고를 더 어렵게 모니터링할 수 있다고 합니다.","url":"https://www.aioga.com/ko/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:43:13.574Z"},"es":{"title":"La nueva tecnología de razonamiento de OpenAI genera preocupación entre los expertos en seguridad de IA: Astra utiliza ciclos opacos que podrían afectar la supervisión del pensamiento en cadena","summary":"Según un informe de The Information, el nuevo modelo Astra de OpenAI utilizará una técnica de razonamiento llamada 'profundidad recurrente' (recurrent depth), que permite al modelo procesar la misma consulta varias veces de manera iterativa, dejando menos rastros legibles, lo que podría dificultar el monitoreo del pensamiento en cadena.","category":"Industria","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"La nueva tecnología de razonamiento de OpenAI genera preocupación entre los expertos en seguridad de IA: Astra utiliza ciclos opacos que podrían afectar la supervisión del pensamiento en cadena - Aioga Noticias de IA","description":"Según un informe de The Information, el nuevo modelo Astra de OpenAI utilizará una técnica de razonamiento llamada 'profundidad recurrente' (recurrent depth), que permite al modelo...","url":"https://www.aioga.com/es/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:43:10.402Z"},"fr":{"title":"La nouvelle technologie de raisonnement d'OpenAI suscite des inquiétudes chez les experts en sécurité de l'IA : Astra utilise des boucles opaques pouvant affecter la surveillance de la pensée en chaîne","summary":"Selon The Information, le nouveau modèle d'OpenAI, Astra, utilisera une technique de raisonnement appelée 'profondeur récurrente' (recurrent depth), permettant au modèle de traiter une même requête de manière répétitive en boucle, en laissant moins de traces lisibles, ce qui pourrait rendre la surveillance de la pensée en chaîne plus difficile.","category":"Industrie","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"La nouvelle technologie de raisonnement d'OpenAI suscite des inquiétudes chez les experts en sécurité de l'IA : Astra utilise des boucles opaques pouvant affecter la surveillance de la pensée en chaîne - Aioga Actualités IA","description":"Selon The Information, le nouveau modèle d'OpenAI, Astra, utilisera une technique de raisonnement appelée 'profondeur récurrente' (recurrent depth), permettant au modèle de traiter...","url":"https://www.aioga.com/fr/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:43:48.037Z"},"de":{"title":"Neue Schlussfolgerungstechnologie von OpenAI löst Sorgen bei KI-Sicherheitsexperten aus: Astra verwendet undurchsichtige Schleifen, die die Überwachung des Ketten-Denkens beeinflussen könnten","summary":"Laut The Information wird das neue Modell Astra von OpenAI eine als recurrent depth (undurchsichtige Schleife) bezeichnete Schlussfolgerungstechnik verwenden, die es dem Modell ermöglicht, dieselbe Abfrage mehrfach zyklisch zu verarbeiten, wobei weniger lesbare Spuren hinterlassen werden, was die Überwachung des Kette-denkens möglicherweise erschwert.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Neue Schlussfolgerungstechnologie von OpenAI löst Sorgen bei KI-Sicherheitsexperten aus: Astra verwendet undurchsichtige Schleifen, die die Überwachung des Ketten-Denkens beeinflussen könnten - Aioga KI-News","description":"Laut The Information wird das neue Modell Astra von OpenAI eine als recurrent depth (undurchsichtige Schleife) bezeichnete Schlussfolgerungstechnik verwenden, die es dem Modell erm...","url":"https://www.aioga.com/de/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:43:53.229Z"},"pt-BR":{"title":"A nova tecnologia de raciocínio da OpenAI desperta preocupações de especialistas em segurança de IA: Astra adota ciclos opacos que podem afetar o monitoramento do pensamento em cadeia","summary":"De acordo com o The Information, o novo modelo Astra da OpenAI utilizará uma técnica de raciocínio chamada recurrent depth (profundidade recorrente), permitindo que o modelo processe a mesma consulta várias vezes de forma cíclica, deixando menos rastros legíveis, o que pode tornar o monitoramento do raciocínio em cadeia mais difícil.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"A nova tecnologia de raciocínio da OpenAI desperta preocupações de especialistas em segurança de IA: Astra adota ciclos opacos que podem afetar o monitoramento do pensamento em cadeia - Aioga Notícias de IA","description":"De acordo com o The Information, o novo modelo Astra da OpenAI utilizará uma técnica de raciocínio chamada recurrent depth (profundidade recorrente), permitindo que o modelo proces...","url":"https://www.aioga.com/pt-BR/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:44:26.423Z"},"ru":{"title":"Новая технология рассуждений OpenAI вызывает обеспокоенность экспертов по безопасности ИИ: Astra использует непрозрачные циклы, что может повлиять на мониторинг цепочечных рассуждений","summary":"Согласно сообщению The Information, новая модель OpenAI Astra будет использовать технологию рассуждений под названием recurrent depth (непрозрачная рекурсия), позволяющую модели многократно обрабатывать один и тот же запрос циклическим образом, оставляя меньше читаемых следов, что может затруднить мониторинг цепочек размышлений.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Новая технология рассуждений OpenAI вызывает обеспокоенность экспертов по безопасности ИИ: Astra использует непрозрачные циклы, что может повлиять на мониторинг цепочечных рассуждений - Aioga Новости ИИ","description":"Согласно сообщению The Information, новая модель OpenAI Astra будет использовать технологию рассуждений под названием recurrent depth (непрозрачная рекурсия), позволяющую модели мн...","url":"https://www.aioga.com/ru/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:44:37.181Z"},"ar":{"title":"تكنولوجيا الاستدلال الجديدة من OpenAI تثير قلق خبراء أمان الذكاء الاصطناعي: Astra تستخدم دورات غير شفافة قد تؤثر على مراقبة التفكير التسلسلي","summary":"وفقًا لتقرير The Information، سيستخدم نموذج OpenAI الجديد Astra تقنية استدلال تُسمى العمق التكراري (دورة غير شفافة)، مما يسمح للنموذج بمعالجة نفس الاستعلام عدة مرات بطريقة دورية، مع ترك آثار مقروءة أقل، مما قد يجعل من الصعب مراقبة التفكير التسلسلي.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"تكنولوجيا الاستدلال الجديدة من OpenAI تثير قلق خبراء أمان الذكاء الاصطناعي: Astra تستخدم دورات غير شفافة قد تؤثر على مراقبة التفكير التسلسلي - Aioga أخبار الذكاء الاصطناعي","description":"وفقًا لتقرير The Information، سيستخدم نموذج OpenAI الجديد Astra تقنية استدلال تُسمى العمق التكراري (دورة غير شفافة)، مما يسمح للنموذج بمعالجة نفس الاستعلام عدة مرات بطريقة دورية، م...","url":"https://www.aioga.com/ar/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:45:17.455Z"},"hi":{"title":"OpenAI की नई तर्क तकनीक ने AI सुरक्षा विशेषज्ञों में चिंता उत्पन्न कर दी: Astra में अपारदर्शी लूप का उपयोग श्रृंखला-आधारित सोच निगरानी को प्रभावित कर सकता है","summary":"The Information की रिपोर्ट के अनुसार, OpenAI का नया मॉडल Astra 'recurrent depth' (अस्पष्ट पुनरावर्ती) नामक तर्क तकनीक का उपयोग करेगा, जिससे मॉडल एक ही प्रश्न को कई बार चक्रीय रूप से संसाधित करेगा, और कम पठनीय निशान छोड़ेगा, जो श्रृंखला-युक्त सोच की निगरानी को और कठिन बना सकता है।","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI की नई तर्क तकनीक ने AI सुरक्षा विशेषज्ञों में चिंता उत्पन्न कर दी: Astra में अपारदर्शी लूप का उपयोग श्रृंखला-आधारित सोच निगरानी को प्रभावित कर सकता है - Aioga AI समाचार","description":"The Information की रिपोर्ट के अनुसार, OpenAI का नया मॉडल Astra 'recurrent depth' (अस्पष्ट पुनरावर्ती) नामक तर्क तकनीक का उपयोग करेगा, जिससे मॉडल एक ही प्रश्न को कई बार चक्रीय रूप स...","url":"https://www.aioga.com/hi/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:45:23.088Z"},"it":{"title":"La nuova tecnologia di ragionamento di OpenAI suscita preoccupazioni negli esperti di sicurezza AI: Astra utilizza cicli opachi che potrebbero influenzare il monitoraggio del pensiero a catena","summary":"Secondo quanto riportato da The Information, il nuovo modello di OpenAI, Astra, utilizzerà una tecnica di ragionamento chiamata recurrent depth (loop opaco), che consente al modello di elaborare più volte la stessa query in modalità ciclica, lasciando meno tracce leggibili, il che potrebbe rendere più difficile monitorare il pensiero a catena.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"La nuova tecnologia di ragionamento di OpenAI suscita preoccupazioni negli esperti di sicurezza AI: Astra utilizza cicli opachi che potrebbero influenzare il monitoraggio del pensiero a catena - Aioga Notizie IA","description":"Secondo quanto riportato da The Information, il nuovo modello di OpenAI, Astra, utilizzerà una tecnica di ragionamento chiamata recurrent depth (loop opaco), che consente al modell...","url":"https://www.aioga.com/it/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:46:01.242Z"},"nl":{"title":"OpenAI's nieuwe redeneertechnologie roept zorgen op bij AI-veiligheidsexperts: Astra maakt gebruik van ondoorzichtige lussen die de monitoring van ketendenken kunnen beïnvloeden","summary":"Volgens The Information zal het nieuwe model Astra van OpenAI een redeneertechniek genaamd recurrent depth (ondoorzichtige herhaling) gebruiken, waardoor het model dezelfde query meerdere keren op een circulaire manier kan verwerken, met minder leesbare sporen achterlatend, wat het mogelijk moeilijker maakt om ketenmatig denken te monitoren.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI's nieuwe redeneertechnologie roept zorgen op bij AI-veiligheidsexperts: Astra maakt gebruik van ondoorzichtige lussen die de monitoring van ketendenken kunnen beïnvloeden - Aioga AI-nieuws","description":"Volgens The Information zal het nieuwe model Astra van OpenAI een redeneertechniek genaamd recurrent depth (ondoorzichtige herhaling) gebruiken, waardoor het model dezelfde query m...","url":"https://www.aioga.com/nl/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:46:02.139Z"},"tr":{"title":"OpenAI'nin yeni akıl yürütme teknolojisi AI güvenlik uzmanlarını endişelendiriyor: Astra'nın şeffaf olmayan döngüleri veya zincirleme düşünce izlemeyi etkilemesi","summary":"The Information'ın haberine göre, OpenAI'nin yeni modeli Astra, modelin aynı sorguyu döngüsel olarak birden çok kez işlemesini sağlayan ve daha az okunabilir iz bırakan 'recurrent depth' (opak döngü) adlı akıl yürütme tekniğini kullanacak ve zincirleme düşünmenin izlenmesini daha zor hale getirebilir.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI'nin yeni akıl yürütme teknolojisi AI güvenlik uzmanlarını endişelendiriyor: Astra'nın şeffaf olmayan döngüleri veya zincirleme düşünce izlemeyi etkilemesi - Aioga AI Haberleri","description":"The Information'ın haberine göre, OpenAI'nin yeni modeli Astra, modelin aynı sorguyu döngüsel olarak birden çok kez işlemesini sağlayan ve daha az okunabilir iz bırakan 'recurrent...","url":"https://www.aioga.com/tr/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:46:42.786Z"},"vi":{"title":"Công nghệ suy luận mới của OpenAI khiến các chuyên gia an ninh AI lo ngại: Astra sử dụng vòng lặp không minh bạch có thể ảnh hưởng đến giám sát tư duy chuỗi","summary":"Theo báo cáo của The Information, mô hình mới Astra của OpenAI sẽ sử dụng kỹ thuật suy luận có tên là recurrent depth (vòng lặp không trong suốt), cho phép mô hình xử lý cùng một truy vấn nhiều lần theo vòng lặp, để lại ít dấu vết có thể đọc được hơn, có thể làm cho việc theo dõi chuỗi suy nghĩ trở nên khó hơn.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Công nghệ suy luận mới của OpenAI khiến các chuyên gia an ninh AI lo ngại: Astra sử dụng vòng lặp không minh bạch có thể ảnh hưởng đến giám sát tư duy chuỗi - Tin tức AI Aioga","description":"Theo báo cáo của The Information, mô hình mới Astra của OpenAI sẽ sử dụng kỹ thuật suy luận có tên là recurrent depth (vòng lặp không trong suốt), cho phép mô hình xử lý cùng một t...","url":"https://www.aioga.com/vi/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:46:38.412Z"},"id":{"title":"Teknologi penalaran baru OpenAI memicu kekhawatiran pakar keamanan AI: Astra menggunakan siklus tidak transparan atau memengaruhi pemantauan pemikiran berantai","summary":"Menurut laporan The Information, model baru OpenAI, Astra, akan menggunakan teknik penalaran yang disebut recurrent depth (siklus tak tembus), yang memungkinkan model memproses kueri yang sama secara berulang, meninggalkan jejak yang lebih sedikit yang dapat dibaca, dan kemungkinan membuat pemantauan pemikiran berantai menjadi lebih sulit.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Teknologi penalaran baru OpenAI memicu kekhawatiran pakar keamanan AI: Astra menggunakan siklus tidak transparan atau memengaruhi pemantauan pemikiran berantai - Berita AI Aioga","description":"Menurut laporan The Information, model baru OpenAI, Astra, akan menggunakan teknik penalaran yang disebut recurrent depth (siklus tak tembus), yang memungkinkan model memproses kue...","url":"https://www.aioga.com/id/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:47:18.322Z"},"th":{"title":"เทคโนโลยีการให้เหตุผลใหม่ของ OpenAI ก่อให้เกิดความกังวลแก่ผู้เชี่ยวชาญด้านความปลอดภัย AI: Astra ใช้วงจรที่ไม่โปร่งใสอาจส่งผลต่อการควบคุมความคิดแบบเป็นลำดับขั้น","summary":"ตามรายงานของ The Information โมเดลใหม่ Astra ของ OpenAI จะใช้เทคโนโลยีการให้เหตุผลที่เรียกว่า recurrent depth (การวนซ้ำที่ไม่โปร่งใส) ซึ่งจะทำให้โมเดลประมวลผลคำถามเดียวกันซ้ำหลายครั้งในลักษณะวนซ้ำ ทิ้งร่องรอยที่อ่านได้น้อยลง และอาจทำให้การตรวจสอบความคิดเชิงลำดับซับซ้อนยากขึ้น","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"เทคโนโลยีการให้เหตุผลใหม่ของ OpenAI ก่อให้เกิดความกังวลแก่ผู้เชี่ยวชาญด้านความปลอดภัย AI: Astra ใช้วงจรที่ไม่โปร่งใสอาจส่งผลต่อการควบคุมความคิดแบบเป็นลำดับขั้น - ข่าว AI Aioga","description":"ตามรายงานของ The Information โมเดลใหม่ Astra ของ OpenAI จะใช้เทคโนโลยีการให้เหตุผลที่เรียกว่า recurrent depth (การวนซ้ำที่ไม่โปร่งใส) ซึ่งจะทำให้โมเดลประมวลผลคำถามเดียวกันซ้ำหลายคร...","url":"https://www.aioga.com/th/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:47:23.367Z"},"pl":{"title":"Nowa technologia wnioskowania OpenAI wywołuje obawy ekspertów ds. bezpieczeństwa AI: Astra używa nieprzejrzystych pętli, co może wpływać na monitorowanie myślenia łańcuchowego","summary":"Według doniesień The Information, nowy model OpenAI o nazwie Astra będzie korzystał z technologii wnioskowania zwanej recurrent depth (nieprzezroczysta pętla), która pozwala modelowi wielokrotnie przetwarzać to samo zapytanie w sposób cykliczny, pozostawiając mniej czytelnych śladów, co może utrudniać monitorowanie myślenia łańcuchowego.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"Nowa technologia wnioskowania OpenAI wywołuje obawy ekspertów ds. bezpieczeństwa AI: Astra używa nieprzejrzystych pętli, co może wpływać na monitorowanie myślenia łańcuchowego - Aioga Wiadomości AI","description":"Według doniesień The Information, nowy model OpenAI o nazwie Astra będzie korzystał z technologii wnioskowania zwanej recurrent depth (nieprzezroczysta pętla), która pozwala modelo...","url":"https://www.aioga.com/pl/news/cmtkjxgun03wqroll8e8qykqg/","contentTranslated":true,"sourceHash":"502f6881d8ed8874","translatedAt":"2026-09-02T20:48:08.192Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}