{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-10-22T20:00:36.154Z","headline":"OpenAI 安全员工 David Robinson 离职，称公司文化已坏","description":"在 OpenAI 工作三年半、负责撰写重大产品发布安全报告的 David Robinson 在 The Atlantic 发文宣布离职，称公司文化已坏。他认为迭代部署式的试错方法必然带来周期性失败，且随系统能力增强风险扩大，主张前沿 AI 公司应像核电站或机场一样以多层冗余运营，并将对齐视为待解的关键问题。","url":"https://www.aioga.com/news/xmlfce496prtvqpqmcupmsssc/","mainEntityOfPage":"https://www.aioga.com/news/xmlfce496prtvqpqmcupmsssc/","datePublished":"2026-10-03T16:30:01.000Z","dateModified":"2026-10-03T16:30:01.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://techcrunch.com/2026/10/03/openai-safety-employee-resigns-claiming-the-companys-culture-is-broken","https://aihot.news/items/xmlfce496prtvqpqmcupmsssc"],"canonicalUrl":"https://www.aioga.com/news/xmlfce496prtvqpqmcupmsssc/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：在 OpenAI 工作三年半、负责撰写重大产品发布安全报告的 David Robinson 在 The Atlantic 发文宣布离职，称公司文化已坏。 Aioga 将其归入「行业动态」方向，重点关注它对真实使用和行业竞争的影响。","url":"https://www.aioga.com/news/xmlfce496prtvqpqmcupmsssc/","dateCreated":"2026-10-03T16:30:01.000Z","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/2026/10/03/openai-safety-employee-resigns-claiming-the-companys-culture-is-broken","datePublished":"2026-10-03T16:30:01.000Z","provider":{"@type":"Organization","name":"TechCrunch","url":"https://techcrunch.com/2026/10/03/openai-safety-employee-resigns-claiming-the-companys-culture-is-broken"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.news/items/xmlfce496prtvqpqmcupmsssc","datePublished":"2026-10-03T16:30:01.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.news/items/xmlfce496prtvqpqmcupmsssc"}}],"aggregationSource":"TechCrunch：AI","originalPublisher":{"name":"TechCrunch","url":"https://techcrunch.com/2026/10/03/openai-safety-employee-resigns-claiming-the-companys-culture-is-broken"},"geoDeepAnswer":null,"article":{"id":"xmlfce496prtvqpqmcupmsssc","slug":"xmlfce496prtvqpqmcupmsssc","url":"https://www.aioga.com/news/xmlfce496prtvqpqmcupmsssc/","title":"OpenAI 安全员工 David Robinson 离职，称公司文化已坏","title_en":"","summary":"在 OpenAI 工作三年半、负责撰写重大产品发布安全报告的 David Robinson 在 The Atlantic 发文宣布离职，称公司文化已坏。他认为迭代部署式的试错方法必然带来周期性失败，且随系统能力增强风险扩大，主张前沿 AI 公司应像核电站或机场一样以多层冗余运营，并将对齐视为待解的关键问题。","source":"TechCrunch：AI","sourceUrl":"https://techcrunch.com/2026/10/03/openai-safety-employee-resigns-claiming-the-companys-culture-is-broken","aiHotUrl":"https://aihot.news/items/xmlfce496prtvqpqmcupmsssc","publishedAt":"2026-10-03T16:30:01.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["By his own admission, David Robinson is “something of a cliché”: an employee at a leading AI company who issues a dire warning while resigning from their job.","In an essay published in The Atlantic：https://www.theatlantic.com/technology/2026/10/openai-safety-team-resignation/688881/?gift=1ga2TvL-DbuHDQIcYF7oR4o908Fsjxr4NFLlsptkfP8, Robinson said he led the writing of safety reports that accompanied OpenAI’s major product launches. He also said that with three-and-a-half years at OpenAI, he is “among the longest-tenured employees at the company.” Now he’s quitting, because in his view, the company’s “culture is broken.”","In some ways, Robinson’s comments echo those of Jacob Coxon, who worked as a researcher at both OpenAI and Anthropic before quitting and declaring that these companies are “gambling with our lives.”：https://techcrunch.com/2026/09/09/gambling-with-our-lives-anthropic-researcher-quits-warns-against-self-improving-ai/ Coxon’s comments led to a broader debate about AI safety, with Anthropic CEO Dario Amodei unveiling a plan for more cautious AI development：https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier/; AI executives met with President Donald Trump this week and signed what appeared to be hastily written, non-binding pledge to implement more safety controls：https://techcrunch.com/2026/09/30/pledge-signed-by-president-trump-and-top-ai-leaders-misspells-the-united-states/.","But in Robinson’s view, the debate needs to go beyond “specific rules or new laws,” addressing the overall culture at these companies. And while much of the reporting around OpenAI has focused on how the company’s CEO Sam Altman lost the trust of former colleagues：https://techcrunch.com/2026/04/11/sam-altman-responds-to-incendiary-new-yorker-article-after-attack-on-his-home/, Robinson’s essay suggests that OpenAI’s culture issues are the same as those of Silicon Valley at large.","“OpenAI has thrived by trial and error (which it calls ‘iterative deployment’), looking for problems and improving its guardrails in response,” he wrote. “But this approach, by its very nature, guarantees periodic failures — and the scale of those failures is growing as systems get more capable.”","Pointing to the recent breach of Hugging Face systems by OpenAI agents, as well as continuing revelations of OpenAI discovering more rogue agents：https://techcrunch.com/2026/09/28/openai-still-doesnt-seem-to-have-a-handle-on-all-of-its-rogue-ai-activity/, Robinson argued, “An environment where things like this can happen is no place to grow artificial minds that could be smarter than we are and that might not do what we want them to.”","Given the increased risk, Robinson argued that frontier AI companies need to start operating “like nuclear-power plants or busy airports, with layers of redundancy and careful, time-consuming planning, so that the occasional and inevitable human error does not open a door to disaster.”","But Robinson said that in his time at OpenAI, he “never encountered a colleague who had experience making airplanes fly safely or nuclear reactors run without melting down, or helping the financial system grow without collapsing.”","In response to Robinson’s essay, OpenAI spokesperson Drew Pusateri said the company continues to improve its safety measures.","“We’re making sure our models don’t become more capable than we can safely manage and secure, and we pause training or hold back models when we need to slow down,” Pusateri said in a statement. “We’re making significant changes to strengthen security in our research and testing environments, train models to not just complete tasks but do so responsibly, expand our work with third-party evaluators, and improve real-time monitoring so we can detect and respond to concerning behavior earlier in the training process：https://alignment.openai.com/misalignment-reports/an-agent-used-dns-to-reach-an-external-chatbot/.”","Beyond calling for changes in OpenAI’s culture, Robinson also said it’s time to ask bigger questions about alignment — something that he admitted could sound “touchy-feely,” but he said it’s critical as companies’ current “measures of how well” AI systems “match human values are coarse.”","“The smarter the industry lets models grow while these problems remain unsolved, the more dangerous our situation becomes,” he said.","Robinson’s departure was first reported by Business Insider：https://www.businessinsider.com/safety-leader-david-robinson-resigns-from-openai-2026-10. In his essay, he also acknowledged that he’s following an apparently a common step in the AI whistleblower playbook: He’s hired a PR firm：https://nypost.com/2026/09/25/business/ai-doomer-pr-firm-worked-for-jacob-coxon-as-he-sparked-furor-over-safety-while-denying-third-party-help-report/. But he insisted, “The decision to speak out is mine alone.”","“Perhaps I should have stayed and fought for fundamental shifts in our staffing and culture, but in practice, my colleagues and I were so busy sprinting that we seldom had the chance to consider big changes, much less to actually make them,” Robinson said. “That’s why I concluded that stronger incentives for safety — coming from outside the company — are a big part of getting this right.”","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.","Anthony Ha is TechCrunch’s weekend editor. Previously, he worked as a tech reporter at Adweek, a senior editor at VentureBeat, a local government reporter at the Hollister Free Lance, and vice president of content at a VC firm. He lives in New York City.","You can contact or verify outreach from Anthony by emailing anthony.ha@techcrunch.com：mailto:anthony.ha@techcrunch.com.","Get 50% off a second pass The Disrupt experience is meant to be shared. Get your pass and bring a colleague, partner, or peer at 50% off. Cover more ground by making connections, building momentum, and discovering what’s next in the startup ecosystem.","Google thinks SpaceX’s Starship has to launch 1,800 times before space data centers get off the ground：https://techcrunch.com/2026/10/01/google-thinks-spacexs-starship-has-to-launch-1600-times-before-space-data-centers-get-off-the-ground/ Tim Fernholz：https://techcrunch.com/author/tim-fernholz/","World’s first enhanced geothermal power plant completed in just 23 months：https://techcrunch.com/2026/10/01/worlds-first-enhanced-geothermal-power-plant-completed-in-just-23-months/ Tim De Chant：https://techcrunch.com/author/tim-de-chant/","Google releases Gemini 4 Argon, called its most powerful model yet：https://techcrunch.com/2026/09/30/google-releases-gemini-4-argon-called-its-most-powerful-model-yet/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/","The Pentagon taps Elon Musk and Palmer Luckey to help decide what the military should do next：https://techcrunch.com/2026/09/30/the-pentagon-taps-elon-musk-and-palmer-luckey-to-help-decide-what-the-military-should-do-next/ Dominic-Madori Davis：https://techcrunch.com/author/dominic-madori-davis/","OpenAI launches Dots, its bubbly agentic avatar：https://techcrunch.com/2026/09/29/openai-launches-dots-its-bubbly-agentic-avatar/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/","AMD will acquire Fei-Fei Li’s World Labs for $8.2B：https://techcrunch.com/2026/09/28/amd-will-acquire-fei-fei-lis-world-labs-for-8-2-billion/ Tim Fernholz：https://techcrunch.com/author/tim-fernholz/","Viral AI agent Instinct raises $1B Series C at a $10B valuation：https://techcrunch.com/2026/09/28/viral-ai-agent-instinct-raises-1b-series-c-at-a-10b-valuation/ Sarah Perez：https://techcrunch.com/author/sarah-perez/"],"articleImages":[{"sourceUrl":"https://techcrunch.com/wp-content/uploads/2025/04/Disrupt2026-Color.png","alt":"Event Logo","afterParagraph":16,"url":"/media/articles/xmlfce496prtvqpqmcupmsssc/d1d06f6f97361f52.webp"}],"mediaStatus":"ok","articleBodyZh":["据他自己承认，大卫·罗宾逊有点“老套”：他是一家领先的人工智能公司的员工，在辞职时发出了严重警告。","罗宾逊在《大西洋》发表的一篇文章中写道：https://www.theatlantic.com/technology/2026/10/openai-safety-team-resignation/688881/?gift=1ga2TvL-DbuHDQIcYF7oR4o908Fsjxr4NFLlsptkfP8，他表示自己主导了OpenAI重大产品发布时的安全报告撰写工作。他还说，自己在OpenAI工作了三年半，是“公司在职时间最长的员工之一”。现在他辞职了，因为在他看来，公司“文化已经破裂”。","在某种程度上，罗宾逊的评论呼应了雅各布·考克森的观点，他曾在OpenAI和Anthropic担任研究员，辞职后宣称这些公司正在“拿我们的生命做赌注”：https://techcrunch.com/2026/09/09/gambling-with-our-lives-anthropic-researcher-quits-warns-against-self-improving-ai/ 考克森的评论引发了关于人工智能安全的更广泛讨论，Anthropic首席执行官达里奥·阿莫代公开了一项更谨慎的人工智能开发计划：https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier/；本周人工智能高管与唐纳德·特朗普总统会面，并签署了一份看似仓促撰写、没有约束力的承诺书，面向具身任务更多安全控制：https://techcrunch.com/2026/09/30/pledge-signed-by-president-trump-and-top-ai-leaders-misspells-the-united-states/。","但在罗宾逊看来，辩论需要超越“具体规则或新法律”，关注这些公司的整体文化。虽然围绕OpenAI的大部分报道集中在公司首席执行官山姆·奥特曼如何失去前同事的信任上：https://techcrunch.com/2026/04/11/sam-altman-responds-to-incendiary-new-yorker-article-after-attack-on-his-home/，罗宾逊的文章表明，OpenAI的文化问题与整个硅谷的状况相同。","“OpenAI通过试错（他们称之为‘迭代部署’）来发展，寻找问题并相应改进其防护措施，”他写道。“但这种方法本质上保证了周期性的失败——随着系统能力的提升，这些失败的规模也在增加。”","罗宾逊指出近期OpenAI代理入侵Hugging Face系统的事件，以及OpenAI持续发现更多流氓代理的情况：https://techcrunch.com/2026/09/28/openai-still-doesnt-seem-to-have-a-handle-on-all-of-its-rogue-ai-activity/，他认为：“一个允许这种事情发生的环境，绝不是培养可能比我们更聪明、并且可能不会按我们期望行事的人工智能的地方。”","鉴于风险增加，罗宾逊认为，前沿AI公司需要开始像运营“核电厂或繁忙机场一样运作，拥有多层冗余和细致耗时的规划，以确保偶尔且不可避免的人为错误不会开启通往灾难的大门。”","但罗宾逊表示，在他任职OpenAI期间，他“从未遇到过有同事具备让飞机安全飞行或核反应堆不熔毁的经验，也没有帮助金融系统在不崩溃的情况下增长的经验”。","针对罗宾逊的文章，OpenAI发言人Drew Pusateri表示，公司正在持续改进其安全措施。","“我们正在确保我们的模型不会变得比我们能安全管理和保护的能力更强，当需要放缓时，我们会暂停训练或推迟发布模型，”Pusateri在一份声明中表示。“我们正在进行重大改进，以加强研究和测试环境中的安全性，训练模型不仅完成任务，还要以负责任的方式执行，扩大与第三方评估员的合作，并改进实时监控，以便在训练过程中更早发现并应对令人担忧的行为：https://alignment.openai.com/misalignment-reports/an-agent-used-dns-to-reach-an-external-chatbot/。”","除了呼吁改变OpenAI的文化之外，罗宾逊还表示，是时候提出更大的对齐问题——他承认这听起来可能“感性”，但他说这是关键，因为公司当前“衡量AI系统与人类价值观匹配程度的方法太粗略”。","“在这些问题未解决的情况下，行业让模型越聪明地成长，我们的处境就越危险，”他说。","罗宾逊的离职最早由《商业内幕》报道：https://www.businessinsider.com/safety-leader-david-robinson-resigns-from-openai-2026-10。在他的文章中，他还承认自己正在遵循人工智能告密者常见的做法：他雇佣了一家公关公司：https://nypost.com/2026/09/25/business/ai-doomer-pr-firm-worked-for-jacob-coxon-as-he-sparked-furor-over-safety-while-denying-third-party-help-report/。但他坚持说，“决定发声完全是我自己的选择。”","“也许我应该留下来，为我们的人员配置和文化进行根本性的改变而奋斗，但实际上，我和我的同事们忙得不可开交，很少有机会去考虑大的变化，更不用说真正去实施它们了，”罗宾逊说。“所以我得出的结论是：来自公司外部的更强安全激励是实现这一目标的重要部分。”","当你通过我们文章中的链接购买时，我们可能会获得少量佣金：https://techcrunch.com/techcrunch-affiliate-monetization-standards/。这不会影响我们的编辑独立性。","Anthony Ha 是 TechCrunch 的周末编辑。之前，他曾在 Adweek 担任科技记者，在 VentureBeat 担任高级编辑，在 Hollister Free Lance 担任地方政府记者，并曾在一家风险投资公司担任内容副总裁。他现居纽约市。","你可以通过发送电子邮件至 anthony.ha@techcrunch.com 与 Anthony 联系或核实工作联系：mailto:anthony.ha@techcrunch.com。","购买第二张票可享五折优惠。Disrupt 活动旨在分享体验。获取你的门票并邀请同事、合作伙伴或同行，享受五折优惠。通过建立联系、积累势能并探索创业生态系统的未来，覆盖更多领域。","谷歌认为，SpaceX 的 Starship 必须发射 1,800 次，空间数据中心才能起步：https://techcrunch.com/2026/10/01/google-thinks-spacexs-starship-has-to-launch-1600-times-before-space-data-centers-get-off-the-ground/。Tim Fernholz：https://techcrunch.com/author/tim-fernholz/","世界首座增强型地热发电厂在短短 23 个月内建成：https://techcrunch.com/2026/10/01/worlds-first-enhanced-geothermal-power-plant-completed-in-just-23-months/。Tim De Chant：https://techcrunch.com/author/tim-de-chant/","谷歌发布Gemini 4 Argon，被称为迄今最强大的模型：https://techcrunch.com/2026/09/30/google-releases-gemini-4-argon-called-its-most-powerful-model-yet/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/","五角大楼邀请埃隆·马斯克和帕尔默·拉基参与决定军方下一步行动：https://techcrunch.com/2026/09/30/the-pentagon-taps-elon-musk-and-palmer-luckey-to-help-decide-what-the-military-should-do-next/ Dominic-Madori Davis：https://techcrunch.com/author/dominic-madori-davis/","OpenAI 推出 Dots，它充满活力的代理化身：https://techcrunch.com/2026/09/29/openai-launches-dots-its-bubbly-agentic-avatar/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/","AMD 将以 82 亿美元收购李飞飞的 World Labs：https://techcrunch.com/2026/09/28/amd-will-acquire-fei-fei-lis-world-labs-for-8-2-billion/ Tim Fernholz：https://techcrunch.com/author/tim-fernholz/","病毒性AI代理Instinct完成10亿美元C轮融资，估值达100亿美元：https://techcrunch.com/2026/09/28/viral-ai-agent-instinct-raises-1b-series-c-at-a-10b-valuation/ Sarah Perez：https://techcrunch.com/author/sarah-perez/"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：在 OpenAI 工作三年半、负责撰写重大产品发布安全报告的 David Robinson 在 The Atlantic 发文宣布离职，称公司文化已坏。 Aioga 将其归入「行业动态」方向，重点关注它对真实使用和行业竞争的影响。","background":"背景分析：公司与行业类动态需要放在竞争格局、商业化路径、资本信号和监管环境中观察，单条公告不能代表最终结果。","viewpoint":"Aioga 判断：这条动态更适合作为行业观察信号，当前信息足以建立线索，但不足以推导长期结论。","implications":"影响分析：对相关团队而言，短期应先核对来源、可用范围和实际成本，再判断是否值得接入或跟进。","nextStep":"后续观察：继续观察官方文件、合作落地、收入或用户信号、竞品动作和监管后续。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-10-22T20:31:24.750Z","sourceHash":"55652798e68f8905","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["行业动态","TechCrunch：AI"],"translations":{"zh-CN":{"title":"OpenAI 安全员工 David Robinson 离职，称公司文化已坏","summary":"在 OpenAI 工作三年半、负责撰写重大产品发布安全报告的 David Robinson 在 The Atlantic 发文宣布离职，称公司文化已坏。他认为迭代部署式的试错方法必然带来周期性失败，且随系统能力增强风险扩大，主张前沿 AI 公司应像核电站或机场一样以多层冗余运营，并将对齐视为待解的关键问题。","category":"行业动态","source":"TechCrunch","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAI 安全员工 David Robinson 离职，称公司文化已坏 - Aioga AI资讯","description":"在 OpenAI 工作三年半、负责撰写重大产品发布安全报告的 David Robinson 在 The Atlantic 发文宣布离职，称公司文化已坏。他认为迭代部署式的试错方法必然带来周期性失败，且随系统能力增强风险扩大，主张前沿 AI 公司应像核电站或机场一样以多层冗余运营，并将对齐视为待解的关键问题。","url":"https://www.aioga.com/news/xmlfce496prtvqpqmcupmsssc/","articleBody":["据他自己承认，大卫·罗宾逊有点“老套”：他是一家领先的人工智能公司的员工，在辞职时发出了严重警告。","罗宾逊在《大西洋》发表的一篇文章中写道：https://www.theatlantic.com/technology/2026/10/openai-safety-team-resignation/688881/?gift=1ga2TvL-DbuHDQIcYF7oR4o908Fsjxr4NFLlsptkfP8，他表示自己主导了OpenAI重大产品发布时的安全报告撰写工作。他还说，自己在OpenAI工作了三年半，是“公司在职时间最长的员工之一”。现在他辞职了，因为在他看来，公司“文化已经破裂”。","在某种程度上，罗宾逊的评论呼应了雅各布·考克森的观点，他曾在OpenAI和Anthropic担任研究员，辞职后宣称这些公司正在“拿我们的生命做赌注”：https://techcrunch.com/2026/09/09/gambling-with-our-lives-anthropic-researcher-quits-warns-against-self-improving-ai/ 考克森的评论引发了关于人工智能安全的更广泛讨论，Anthropic首席执行官达里奥·阿莫代公开了一项更谨慎的人工智能开发计划：https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier/；本周人工智能高管与唐纳德·特朗普总统会面，并签署了一份看似仓促撰写、没有约束力的承诺书，面向具身任务更多安全控制：https://techcrunch.com/2026/09/30/pledge-signed-by-president-trump-and-top-ai-leaders-misspells-the-united-states/。","但在罗宾逊看来，辩论需要超越“具体规则或新法律”，关注这些公司的整体文化。虽然围绕OpenAI的大部分报道集中在公司首席执行官山姆·奥特曼如何失去前同事的信任上：https://techcrunch.com/2026/04/11/sam-altman-responds-to-incendiary-new-yorker-article-after-attack-on-his-home/，罗宾逊的文章表明，OpenAI的文化问题与整个硅谷的状况相同。","“OpenAI通过试错（他们称之为‘迭代部署’）来发展，寻找问题并相应改进其防护措施，”他写道。“但这种方法本质上保证了周期性的失败——随着系统能力的提升，这些失败的规模也在增加。”","罗宾逊指出近期OpenAI代理入侵Hugging Face系统的事件，以及OpenAI持续发现更多流氓代理的情况：https://techcrunch.com/2026/09/28/openai-still-doesnt-seem-to-have-a-handle-on-all-of-its-rogue-ai-activity/，他认为：“一个允许这种事情发生的环境，绝不是培养可能比我们更聪明、并且可能不会按我们期望行事的人工智能的地方。”","鉴于风险增加，罗宾逊认为，前沿AI公司需要开始像运营“核电厂或繁忙机场一样运作，拥有多层冗余和细致耗时的规划，以确保偶尔且不可避免的人为错误不会开启通往灾难的大门。”","但罗宾逊表示，在他任职OpenAI期间，他“从未遇到过有同事具备让飞机安全飞行或核反应堆不熔毁的经验，也没有帮助金融系统在不崩溃的情况下增长的经验”。","针对罗宾逊的文章，OpenAI发言人Drew Pusateri表示，公司正在持续改进其安全措施。","“我们正在确保我们的模型不会变得比我们能安全管理和保护的能力更强，当需要放缓时，我们会暂停训练或推迟发布模型，”Pusateri在一份声明中表示。“我们正在进行重大改进，以加强研究和测试环境中的安全性，训练模型不仅完成任务，还要以负责任的方式执行，扩大与第三方评估员的合作，并改进实时监控，以便在训练过程中更早发现并应对令人担忧的行为：https://alignment.openai.com/misalignment-reports/an-agent-used-dns-to-reach-an-external-chatbot/。”","除了呼吁改变OpenAI的文化之外，罗宾逊还表示，是时候提出更大的对齐问题——他承认这听起来可能“感性”，但他说这是关键，因为公司当前“衡量AI系统与人类价值观匹配程度的方法太粗略”。","“在这些问题未解决的情况下，行业让模型越聪明地成长，我们的处境就越危险，”他说。","罗宾逊的离职最早由《商业内幕》报道：https://www.businessinsider.com/safety-leader-david-robinson-resigns-from-openai-2026-10。在他的文章中，他还承认自己正在遵循人工智能告密者常见的做法：他雇佣了一家公关公司：https://nypost.com/2026/09/25/business/ai-doomer-pr-firm-worked-for-jacob-coxon-as-he-sparked-furor-over-safety-while-denying-third-party-help-report/。但他坚持说，“决定发声完全是我自己的选择。”","“也许我应该留下来，为我们的人员配置和文化进行根本性的改变而奋斗，但实际上，我和我的同事们忙得不可开交，很少有机会去考虑大的变化，更不用说真正去实施它们了，”罗宾逊说。“所以我得出的结论是：来自公司外部的更强安全激励是实现这一目标的重要部分。”","当你通过我们文章中的链接购买时，我们可能会获得少量佣金：https://techcrunch.com/techcrunch-affiliate-monetization-standards/。这不会影响我们的编辑独立性。","Anthony Ha 是 TechCrunch 的周末编辑。之前，他曾在 Adweek 担任科技记者，在 VentureBeat 担任高级编辑，在 Hollister Free Lance 担任地方政府记者，并曾在一家风险投资公司担任内容副总裁。他现居纽约市。","你可以通过发送电子邮件至 anthony.ha@techcrunch.com 与 Anthony 联系或核实工作联系：mailto:anthony.ha@techcrunch.com。","购买第二张票可享五折优惠。Disrupt 活动旨在分享体验。获取你的门票并邀请同事、合作伙伴或同行，享受五折优惠。通过建立联系、积累势能并探索创业生态系统的未来，覆盖更多领域。","谷歌认为，SpaceX 的 Starship 必须发射 1,800 次，空间数据中心才能起步：https://techcrunch.com/2026/10/01/google-thinks-spacexs-starship-has-to-launch-1600-times-before-space-data-centers-get-off-the-ground/。Tim Fernholz：https://techcrunch.com/author/tim-fernholz/","世界首座增强型地热发电厂在短短 23 个月内建成：https://techcrunch.com/2026/10/01/worlds-first-enhanced-geothermal-power-plant-completed-in-just-23-months/。Tim De Chant：https://techcrunch.com/author/tim-de-chant/","谷歌发布Gemini 4 Argon，被称为迄今最强大的模型：https://techcrunch.com/2026/09/30/google-releases-gemini-4-argon-called-its-most-powerful-model-yet/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/","五角大楼邀请埃隆·马斯克和帕尔默·拉基参与决定军方下一步行动：https://techcrunch.com/2026/09/30/the-pentagon-taps-elon-musk-and-palmer-luckey-to-help-decide-what-the-military-should-do-next/ Dominic-Madori Davis：https://techcrunch.com/author/dominic-madori-davis/","OpenAI 推出 Dots，它充满活力的代理化身：https://techcrunch.com/2026/09/29/openai-launches-dots-its-bubbly-agentic-avatar/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/","AMD 将以 82 亿美元收购李飞飞的 World Labs：https://techcrunch.com/2026/09/28/amd-will-acquire-fei-fei-lis-world-labs-for-8-2-billion/ Tim Fernholz：https://techcrunch.com/author/tim-fernholz/","病毒性AI代理Instinct完成10亿美元C轮融资，估值达100亿美元：https://techcrunch.com/2026/09/28/viral-ai-agent-instinct-raises-1b-series-c-at-a-10b-valuation/ Sarah Perez：https://techcrunch.com/author/sarah-perez/"]},"en":{"title":"OpenAI safety employee David Robinson leaves, saying the company culture is broken","summary":"David Robinson, who worked at OpenAI for three and a half years and was responsible for writing safety reports for major product launches, announced his departure in an article published in The Atlantic, saying that the company culture had broken down. He believes that an iterative, deployment-based trial-and-error approach inevitably leads to periodic failures, with risks growing as system capabilities increase. He argues that frontier AI companies should operate with multiple layers of redundancy, like nuclear power plants or airports, and treat alignment as a key unresolved problem.","category":"Industry","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAI safety employee David Robinson leaves, saying the company culture is broken - Aioga AI News","description":"David Robinson, who worked at OpenAI for three and a half years and was responsible for writing safety reports for major product launches, announced his departure in an article pub...","url":"https://www.aioga.com/en/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:08:09.367Z"},"ja":{"title":"OpenAIの安全担当社員、David Robinson氏が退職し、「会社の文化は壊れている」と語る","summary":"OpenAIで3年半働き、重大な製品リリースの安全レポートの作成を担当していたデイビッド・ロビンソンは、The Atlanticに寄稿し、会社を辞めることを発表し、同社の文化は腐敗していると述べた。彼は、反復的なデプロイによる試行錯誤の方法は必ず周期的な失敗をもたらし、システム能力が高まるにつれてリスクが拡大すると考えており、先端AI企業は原子力発電所や空港のように多層の冗長性で運営されるべきだと主張し、アラインメントを解決すべき重要な問題と見なしている。","category":"業界動向","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAIの安全担当社員、David Robinson氏が退職し、「会社の文化は壊れている」と語る - Aioga AIニュース","description":"OpenAIで3年半働き、重大な製品リリースの安全レポートの作成を担当していたデイビッド・ロビンソンは、The Atlanticに寄稿し、会社を辞めることを発表し、同社の文化は腐敗していると述べた。彼は、反復的なデプロイによる試行錯誤の方法は必ず周期的な失敗をもたらし、システム能力が高まるにつれてリスクが拡大すると考えており、先端AI企業は原子力発電所や空港...","url":"https://www.aioga.com/ja/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:09:18.716Z"},"ko":{"title":"OpenAI 안전 직원 David Robinson 퇴직, 회사 문화가 이미 나빠졌다고 언급","summary":"OpenAI에서 3년 반 동안 근무하며 주요 제품 출시 안전 보고서를 작성해온 David Robinson이 The Atlantic에 글을 올려 퇴사를 발표하며 회사 문화가 망가졌다고 말했다. 그는 반복적 배포식 시행착오 방식이 필연적으로 주기적인 실패를 초래하며, 시스템 능력이 향상됨에 따라 위험이 확대된다고 생각하고, 최첨단 AI 회사는 원자력 발전소나 공항처럼 다층 중복 운영을 해야 하며, 정렬 문제를 해결해야 할 핵심 문제로 봐야 한다고 주장한다.","category":"업계 동향","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAI 안전 직원 David Robinson 퇴직, 회사 문화가 이미 나빠졌다고 언급 - Aioga AI 뉴스","description":"OpenAI에서 3년 반 동안 근무하며 주요 제품 출시 안전 보고서를 작성해온 David Robinson이 The Atlantic에 글을 올려 퇴사를 발표하며 회사 문화가 망가졌다고 말했다. 그는 반복적 배포식 시행착오 방식이 필연적으로 주기적인 실패를 초래하며, 시스템 능력이 향상됨에 따라 위험이 확대된다고 생각하고,...","url":"https://www.aioga.com/ko/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:08:57.065Z"},"es":{"title":"El empleado de seguridad de OpenAI, David Robinson, deja la empresa y afirma que la cultura corporativa se ha deteriorado.","summary":"David Robinson, quien trabajó tres años y medio en OpenAI y se encargaba de redactar informes de seguridad sobre importantes lanzamientos de productos, anunció su salida de la empresa en un artículo publicado en The Atlantic, afirmando que la cultura de la compañía se había deteriorado. Considera que el método de prueba y error basado en despliegues iterativos conduce inevitablemente a fallos periódicos y que los riesgos aumentan a medida que los sistemas se vuelven más capaces. Sostiene que las empresas de IA de frontera deberían operar con múltiples capas de redundancia, como las centrales nucleares o los aeropuertos, y considerar la alineación como una cuestión clave aún por resolver.","category":"Industria","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"El empleado de seguridad de OpenAI, David Robinson, deja la empresa y afirma que la cultura corporativa se ha deteriorado. - Aioga Noticias de IA","description":"David Robinson, quien trabajó tres años y medio en OpenAI y se encargaba de redactar informes de seguridad sobre importantes lanzamientos de productos, anunció su salida de la empr...","url":"https://www.aioga.com/es/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:14:20.792Z"},"fr":{"title":"David Robinson, employé chargé de la sécurité chez OpenAI, démissionne, affirmant que la culture de l’entreprise s’est détériorée.","summary":"David Robinson, qui a travaillé pendant trois ans et demi chez OpenAI et était chargé de rédiger les rapports de sécurité sur les lancements de produits majeurs, a annoncé son départ dans un article publié par The Atlantic, affirmant que la culture de l’entreprise s’était détériorée. Il estime que la méthode d’essais par déploiements itératifs entraîne inévitablement des échecs périodiques et que les risques s’accroissent à mesure que les capacités des systèmes augmentent. Il soutient que les entreprises d’IA de pointe devraient fonctionner avec plusieurs niveaux de redondance, à l’instar des centrales nucléaires ou des aéroports, et considérer l’alignement comme un problème crucial restant à résoudre.","category":"Industrie","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"David Robinson, employé chargé de la sécurité chez OpenAI, démissionne, affirmant que la culture de l’entreprise s’est détériorée. - Aioga Actualités IA","description":"David Robinson, qui a travaillé pendant trois ans et demi chez OpenAI et était chargé de rédiger les rapports de sécurité sur les lancements de produits majeurs, a annoncé son dépa...","url":"https://www.aioga.com/fr/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:14:31.729Z"},"de":{"title":"OpenAI-Sicherheitsmitarbeiter David Robinson tritt zurück und sagt, dass die Unternehmenskultur kaputt sei","summary":"David Robinson, der dreieinhalb Jahre bei OpenAI tätig war und für das Verfassen von Sicherheitsberichten zu bedeutenden Produktveröffentlichungen verantwortlich war, gab in einem Beitrag in The Atlantic seinen Rücktritt bekannt und erklärte, die Unternehmenskultur sei verfallen. Seiner Ansicht nach führt die Methode des iterativen Ausrollens und Ausprobierens zwangsläufig zu periodischen Fehlschlägen, wobei die Risiken mit zunehmenden Systemfähigkeiten größer werden. Er plädiert dafür, dass Unternehmen im Bereich der Spitzen-KI mit mehreren Redundanzebenen betrieben werden sollten, ähnlich wie Kernkraftwerke oder Flughäfen, und dass Alignment als eine zentrale noch zu lösende Frage betrachtet werden sollte.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAI-Sicherheitsmitarbeiter David Robinson tritt zurück und sagt, dass die Unternehmenskultur kaputt sei - Aioga KI-News","description":"David Robinson, der dreieinhalb Jahre bei OpenAI tätig war und für das Verfassen von Sicherheitsberichten zu bedeutenden Produktveröffentlichungen verantwortlich war, gab in einem...","url":"https://www.aioga.com/de/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:14:40.282Z"},"pt-BR":{"title":"David Robinson, funcionário da área de segurança da OpenAI, deixa a empresa e afirma que a cultura corporativa se deteriorou","summary":"David Robinson, que trabalhou na OpenAI por três anos e meio e era responsável por redigir relatórios de segurança para grandes lançamentos de produtos, anunciou sua saída em um artigo na The Atlantic, afirmando que a cultura da empresa estava deteriorada. Ele acredita que o método de tentativa e erro com implantação iterativa inevitavelmente leva a falhas cíclicas e que os riscos aumentam à medida que a capacidade do sistema cresce, defendendo que empresas de IA de ponta deveriam operar com múltiplas camadas de redundância, como usinas nucleares ou aeroportos, e considerar o alinhamento como um problema crucial a ser resolvido.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"David Robinson, funcionário da área de segurança da OpenAI, deixa a empresa e afirma que a cultura corporativa se deteriorou - Aioga Notícias de IA","description":"David Robinson, que trabalhou na OpenAI por três anos e meio e era responsável por redigir relatórios de segurança para grandes lançamentos de produtos, anunciou sua saída em um ar...","url":"https://www.aioga.com/pt-BR/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:20:22.884Z"},"ru":{"title":"Сотрудник службы безопасности OpenAI Дэвид Робинсон уходит с работы, заявив, что культура компании испортилась","summary":"Дэвид Робинсон, проработавший в OpenAI три с половиной года и отвечавший за составление отчетов о безопасности при крупных релизах продуктов, объявил о своем уходе в статье для The Atlantic, заявив, что корпоративная культура испорчена. Он считает, что метод итеративного развертывания с пробами и ошибками неизбежно приводит к периодическим сбоям, а с ростом возможностей системы растет и риск, и выступает за то, чтобы ведущие AI-компании работали с многослойным резервированием, как атомные электростанции или аэропорты, рассматривая согласование как ключевую проблему, требующую решения.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"Сотрудник службы безопасности OpenAI Дэвид Робинсон уходит с работы, заявив, что культура компании испортилась - Aioga Новости ИИ","description":"Дэвид Робинсон, проработавший в OpenAI три с половиной года и отвечавший за составление отчетов о безопасности при крупных релизах продуктов, объявил о своем уходе в статье для The...","url":"https://www.aioga.com/ru/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:20:22.369Z"},"ar":{"title":"استقال موظف السلامة في OpenAI، ديفيد روبنسون، قائلاً إن ثقافة الشركة قد فسدت","summary":"ديفيد روبنسون، الذي عمل في OpenAI لمدة ثلاث سنوات ونصف وكان مسؤولًا عن كتابة تقارير الأمان لإصدارات المنتجات الكبيرة، أعلن عن استقالته في مقال نشرته مجلة The Atlantic، قائلًا إن ثقافة الشركة قد فسدت. ويعتقد أن طريقة التجربة والخطأ التي تتبع النشر التدريجي تؤدي حتمًا إلى الفشل الدوري، وأن المخاطر تتزايد مع تعزيز قدرة النظام، ويدعو إلى أن تعمل شركات الذكاء الاصطناعي المتقدمة مثل محطات الطاقة النووية أو المطارات مع تكرار متعدد للسلامة، وأن يُنظر إلى التوجيه باعتباره مشكلة رئيسية تحتاج إلى حل.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"استقال موظف السلامة في OpenAI، ديفيد روبنسون، قائلاً إن ثقافة الشركة قد فسدت - Aioga أخبار الذكاء الاصطناعي","description":"ديفيد روبنسون، الذي عمل في OpenAI لمدة ثلاث سنوات ونصف وكان مسؤولًا عن كتابة تقارير الأمان لإصدارات المنتجات الكبيرة، أعلن عن استقالته في مقال نشرته مجلة The Atlantic، قائلًا إن ثق...","url":"https://www.aioga.com/ar/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:20:09.395Z"},"hi":{"title":"OpenAI के सुरक्षा कर्मचारी डेविड रॉबिंसन ने इस्तीफा दिया, कहा कि कंपनी की संस्कृति खराब हो गई है","summary":"OpenAI में साढ़े तीन साल तक काम करने वाले और प्रमुख उत्पाद लॉन्च की सुरक्षा रिपोर्टें लिखने के लिए जिम्मेदार David Robinson ने The Atlantic में एक लेख प्रकाशित कर अपने इस्तीफे की घोषणा की और कहा कि कंपनी की संस्कृति खराब हो चुकी है। उनका मानना है कि क्रमिक परिनियोजन के माध्यम से परीक्षण-त्रुटि की पद्धति अनिवार्य रूप से समय-समय पर विफलताएं लाएगी, और जैसे-जैसे प्रणालियों की क्षमताएं बढ़ेंगी, जोखिम भी बढ़ता जाएगा। उनका तर्क है कि अग्रणी AI कंपनियों को परमाणु ऊर्जा संयंत्रों या हवाई अड्डों की तरह कई स्तरों वाली अतिरिक्त सुरक्षा के साथ काम करना चाहिए और संरेखण को हल किए जाने वाले एक महत्वपूर्ण प्रश्न के रूप में देखना चाहिए।","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAI के सुरक्षा कर्मचारी डेविड रॉबिंसन ने इस्तीफा दिया, कहा कि कंपनी की संस्कृति खराब हो गई है - Aioga AI समाचार","description":"OpenAI में साढ़े तीन साल तक काम करने वाले और प्रमुख उत्पाद लॉन्च की सुरक्षा रिपोर्टें लिखने के लिए जिम्मेदार David Robinson ने The Atlantic में एक लेख प्रकाशित कर अपने इस्तीफे की घ...","url":"https://www.aioga.com/hi/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:25:39.931Z"},"it":{"title":"David Robinson, dipendente della sicurezza di OpenAI, si dimette, affermando che la cultura aziendale è deteriorata","summary":"David Robinson, che ha lavorato in OpenAI per tre anni e mezzo e si è occupato di redigere rapporti sulla sicurezza per importanti lanci di prodotti, ha annunciato le dimissioni in un articolo su The Atlantic, affermando che la cultura aziendale si è deteriorata. Egli ritiene che il metodo iterativo di dispiegamento e sperimentazione inevitabilmente porti a fallimenti ciclici e che, con l'aumento delle capacità dei sistemi, il rischio cresca. Sostiene che le aziende AI all'avanguardia dovrebbero operare con multi-livello di ridondanza come centrali nucleari o aeroporti, e considerare l'allineamento come una questione chiave ancora da risolvere.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"David Robinson, dipendente della sicurezza di OpenAI, si dimette, affermando che la cultura aziendale è deteriorata - Aioga Notizie IA","description":"David Robinson, che ha lavorato in OpenAI per tre anni e mezzo e si è occupato di redigere rapporti sulla sicurezza per importanti lanci di prodotti, ha annunciato le dimissioni in...","url":"https://www.aioga.com/it/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:25:21.248Z"},"nl":{"title":"OpenAI-veiligheidsmedewerker David Robinson vertrekt en zegt dat de bedrijfscultuur is verziekt","summary":"David Robinson, die drieënhalf jaar bij OpenAI heeft gewerkt en verantwoordelijk was voor het opstellen van veiligheidsrapporten voor belangrijke productlanceringen, kondigde in The Atlantic zijn vertrek aan en zei dat de bedrijfscultuur slecht was. Hij denkt dat de iteratieve implementatie van een proef-en-foutbenadering onvermijdelijk zal leiden tot cyclische mislukkingen, en dat de risico's toenemen naarmate de systeemcapaciteiten groeien. Hij pleit ervoor dat vooraanstaande AI-bedrijven zoals kerncentrales of luchthavens met meervoudige redundantie zouden moeten opereren, en ziet afstemming als een belangrijk probleem dat moet worden opgelost.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAI-veiligheidsmedewerker David Robinson vertrekt en zegt dat de bedrijfscultuur is verziekt - Aioga AI-nieuws","description":"David Robinson, die drieënhalf jaar bij OpenAI heeft gewerkt en verantwoordelijk was voor het opstellen van veiligheidsrapporten voor belangrijke productlanceringen, kondigde in Th...","url":"https://www.aioga.com/nl/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:25:53.964Z"},"tr":{"title":"OpenAI güvenlik çalışanı David Robinson istifa etti, şirket kültürünün bozulduğunu söyledi","summary":"OpenAI'de üç buçuk yıl çalışmış ve önemli ürün lansmanlarının güvenlik raporlarını hazırlamaktan sorumlu olan David Robinson, The Atlantic'te bir makale yayınlayarak istifa ettiğini açıkladı ve şirket kültürünün bozulduğunu söyledi. Robinson, tekrarlamalı dağıtım yöntemiyle yapılan deneme-yanılma sürecinin kaçınılmaz olarak döngüsel başarısızlıklara yol açacağını ve sistem yetenekleri arttıkça riskin büyüyeceğini düşünüyor; öncü yapay zeka şirketlerinin nükleer santral veya havalimanları gibi çok katmanlı yedeklemelerle çalışması gerektiğini savunuyor ve hizalamanın çözülmesi gereken kritik bir sorun olarak görülmesi gerektiğini belirtiyor.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"OpenAI güvenlik çalışanı David Robinson istifa etti, şirket kültürünün bozulduğunu söyledi - Aioga AI Haberleri","description":"OpenAI'de üç buçuk yıl çalışmış ve önemli ürün lansmanlarının güvenlik raporlarını hazırlamaktan sorumlu olan David Robinson, The Atlantic'te bir makale yayınlayarak istifa ettiğin...","url":"https://www.aioga.com/tr/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:31:15.577Z"},"vi":{"title":"Nhân viên an ninh OpenAI David Robinson nghỉ việc, cho biết văn hóa công ty đã hỏng","summary":"David Robinson, người đã làm việc tại OpenAI ba năm rưỡi và chịu trách nhiệm viết báo cáo an toàn cho các lần ra mắt sản phẩm quan trọng, đã đăng bài trên The Atlantic thông báo nghỉ việc, cho biết văn hóa công ty đã xấu đi. Anh cho rằng phương pháp thử nghiệm liên tục theo từng bước chắc chắn sẽ dẫn đến thất bại định kỳ, và khi khả năng hệ thống tăng lên thì rủi ro cũng tăng, đồng thời đề xuất các công ty AI tiên tiến nên vận hành với nhiều lớp dự phòng như nhà máy điện hạt nhân hoặc sân bay, và xem việc đồng bộ hóa là vấn đề quan trọng cần giải quyết.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"Nhân viên an ninh OpenAI David Robinson nghỉ việc, cho biết văn hóa công ty đã hỏng - Tin tức AI Aioga","description":"David Robinson, người đã làm việc tại OpenAI ba năm rưỡi và chịu trách nhiệm viết báo cáo an toàn cho các lần ra mắt sản phẩm quan trọng, đã đăng bài trên The Atlantic thông báo ng...","url":"https://www.aioga.com/vi/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:31:01.111Z"},"id":{"title":"Karyawan keamanan OpenAI David Robinson mengundurkan diri, menyebut budaya perusahaan telah rusak","summary":"David Robinson, yang telah bekerja di OpenAI selama tiga setengah tahun dan bertanggung jawab menulis laporan keamanan untuk peluncuran produk besar, mengumumkan pengunduran dirinya di The Atlantic, menyatakan bahwa budaya perusahaan telah rusak. Ia berpendapat bahwa metode percobaan dan kesalahan berbasis iterasi pasti akan menyebabkan kegagalan berkala, dan risiko meningkat seiring kemampuan sistem semakin kuat. Ia menekankan bahwa perusahaan AI terdepan seharusnya beroperasi dengan redundansi berlapis seperti pembangkit listrik nuklir atau bandara, dan melihat penyelarasan sebagai masalah kunci yang harus dipecahkan.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"Karyawan keamanan OpenAI David Robinson mengundurkan diri, menyebut budaya perusahaan telah rusak - Berita AI Aioga","description":"David Robinson, yang telah bekerja di OpenAI selama tiga setengah tahun dan bertanggung jawab menulis laporan keamanan untuk peluncuran produk besar, mengumumkan pengunduran diriny...","url":"https://www.aioga.com/id/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:31:01.316Z"},"th":{"title":"David Robinson พนักงานฝ่ายความปลอดภัยของ OpenAI ลาออก โดยระบุว่าวัฒนธรรมองค์กรเสื่อมถอยแล้ว","summary":"David Robinson ซึ่งทำงานที่ OpenAI เป็นเวลาสามปีครึ่งและรับผิดชอบการเขียนรายงานด้านความปลอดภัยสำหรับการเปิดตัวผลิตภัณฑ์ครั้งสำคัญ ได้เขียนบทความใน The Atlantic ประกาศลาออก โดยระบุว่าวัฒนธรรมองค์กรได้เสื่อมถอยลง เขาเห็นว่าวิธีการลองผิดลองถูกผ่านการปรับปรุงและการนำระบบไปใช้งานเป็นระยะย่อมก่อให้เกิดความล้มเหลวเป็นวัฏจักร และเมื่อความสามารถของระบบเพิ่มขึ้น ความเสี่ยงก็จะขยายตัวตามไปด้วย เขาเสนอว่าบริษัท AI แนวหน้าควรดำเนินงานด้วยระบบสำรองหลายชั้นเช่นเดียวกับโรงไฟฟ้านิวเคลียร์หรือสนามบิน และควรมองว่าการจัดแนว (alignment) เป็นประเด็นสำคัญที่ยังต้องแก้ไข","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"David Robinson พนักงานฝ่ายความปลอดภัยของ OpenAI ลาออก โดยระบุว่าวัฒนธรรมองค์กรเสื่อมถอยแล้ว - ข่าว AI Aioga","description":"David Robinson ซึ่งทำงานที่ OpenAI เป็นเวลาสามปีครึ่งและรับผิดชอบการเขียนรายงานด้านความปลอดภัยสำหรับการเปิดตัวผลิตภัณฑ์ครั้งสำคัญ ได้เขียนบทความใน The Atlantic ประกาศลาออก โดยระบุว...","url":"https://www.aioga.com/th/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:36:41.990Z"},"pl":{"title":"David Robinson, pracownik ds. bezpieczeństwa w OpenAI, odchodzi z pracy, twierdząc, że kultura firmy jest zepsuta","summary":"David Robinson, który przez trzy i pół roku pracował w OpenAI i odpowiadał za pisanie raportów dotyczących bezpieczeństwa przy dużych premierach produktów, opublikował w The Atlantic artykuł, w którym ogłosił swoje odejście z firmy, twierdząc, że jej kultura uległa zepsuciu. Uważa, że metoda prób i błędów oparta na iteracyjnym wdrażaniu nieuchronnie prowadzi do okresowych porażek, a wraz ze wzrostem możliwości systemów rośnie również ryzyko. Opowiada się za tym, aby firmy rozwijające przełomową sztuczną inteligencję działały z zastosowaniem wielowarstwowej redundancji, podobnie jak elektrownie jądrowe czy lotniska, oraz traktowały alignment jako kluczowy problem wymagający rozwiązania.","category":"行业动态","source":"TechCrunch：AI","aggregationSource":"TechCrunch：AI","pageTitle":"David Robinson, pracownik ds. bezpieczeństwa w OpenAI, odchodzi z pracy, twierdząc, że kultura firmy jest zepsuta - Aioga Wiadomości AI","description":"David Robinson, który przez trzy i pół roku pracował w OpenAI i odpowiadał za pisanie raportów dotyczących bezpieczeństwa przy dużych premierach produktów, opublikował w The Atlant...","url":"https://www.aioga.com/pl/news/xmlfce496prtvqpqmcupmsssc/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"4dc7b529fca6c5c1","translatedAt":"2026-10-03T18:36:11.452Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}