{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-10-23T06:00:40.283Z","headline":"前 OpenAI 安全员工 David Robinson 发文批评公司迭代部署过于激进、文化已崩坏","description":"据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 依赖先发布后加固的迭代部署模式，认为应像核电与航空业那样建立防护机制。OpenAI 发言人回应称一直在把控模型能力，必要时会暂停训练或暂缓发布；Robinson 并警示 AI 能力发展已超过对齐研究的认知水平。","url":"https://www.aioga.com/news/mqv2fnovlxtlcs8sgvw2mz47a/","mainEntityOfPage":"https://www.aioga.com/news/mqv2fnovlxtlcs8sgvw2mz47a/","datePublished":"2026-10-03T22:43:56.000Z","dateModified":"2026-10-03T22:43:56.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://www.ithome.com/1/009/581.htm","https://aihot.news/items/mqv2fnovlxtlcs8sgvw2mz47a"],"canonicalUrl":"https://www.aioga.com/news/mqv2fnovlxtlcs8sgvw2mz47a/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 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/mqv2fnovlxtlcs8sgvw2mz47a/","dateCreated":"2026-10-03T22:43:56.000Z","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":"IT之家 source article","url":"https://www.ithome.com/1/009/581.htm","datePublished":"2026-10-03T22:43:56.000Z","provider":{"@type":"Organization","name":"IT之家","url":"https://www.ithome.com/1/009/581.htm"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.news/items/mqv2fnovlxtlcs8sgvw2mz47a","datePublished":"2026-10-03T22:43:56.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.news/items/mqv2fnovlxtlcs8sgvw2mz47a"}}],"aggregationSource":"IT之家","originalPublisher":{"name":"IT之家","url":"https://www.ithome.com/1/009/581.htm"},"geoDeepAnswer":null,"article":{"id":"mqv2fnovlxtlcs8sgvw2mz47a","slug":"mqv2fnovlxtlcs8sgvw2mz47a","url":"https://www.aioga.com/news/mqv2fnovlxtlcs8sgvw2mz47a/","title":"前 OpenAI 安全员工 David Robinson 发文批评公司迭代部署过于激进、文化已崩坏","title_en":"","summary":"据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 依赖先发布后加固的迭代部署模式，认为应像核电与航空业那样建立防护机制。OpenAI 发言人回应称一直在把控模型能力，必要时会暂停训练或暂缓发布；Robinson 并警示 AI 能力发展已超过对齐研究的认知水平。","source":"IT之家","sourceUrl":"https://www.ithome.com/1/009/581.htm","aiHotUrl":"https://aihot.news/items/mqv2fnovlxtlcs8sgvw2mz47a","publishedAt":"2026-10-03T22:43:56.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["IT之家：https://www.ithome.com/ 10 月 4 日消息，据路透社报道，一名近期辞职的前 OpenAI 安全部门员工批评了该公司对待人工智能安全的做法，他认为一味追求快速开发的高压企业文化会提高系统发生故障的风险。","IT之家注意到，戴维 · 鲁宾森（David Robinson）于当地时间周六在《大西洋》杂志发表题为《我辞去 OpenAI 的工作，只因这家公司的文化已经崩坏》的文章中表示，包括 OpenAI 在内的各家人工智能企业做得“远远不够谨慎”，在研发能力更强的系统之前，应当更加重视安全方面的专业人才与相关研究。","“试错的时代已经结束了。”鲁宾森（David Robinson）写道。他提出，先进人工智能系统所需要的防护机制，应当更接近核电与航空这类行业所采用的安全保障体系。","鲁宾森（David Robinson）称，OpenAI 十分依赖其所谓的“迭代部署”模式：先对外发布系统，等到问题暴露之后再去加固安全防护。","此番言论进一步激化了人工智能行业内部的一场争论，争议焦点在于各家企业研发性能日益强大的系统时，推进的脚步是否过快。此前已经发生过多起安全管控失效或是实验性系统出现异常表现的事件，OpenAI 及其竞争对手 Anthropic 也因此受到外界审视。","鲁宾森自称曾在 OpenAI 任职三年半，参与起草过该公司的风险应对框架，并且负责监督 12 次前沿模型发布对应的安全报告。他写道：“公司只顾着接二连三地赶进度推出新版本，却没能达到我心目中理应具备的审慎标准。”","OpenAI 的一名发言人在一份声明中表示：“我们一直在把控模型，避免它的能力超出我们能够安全管控的范围；一旦需要放缓节奏，我们就会暂停训练，或是暂缓模型的对外发布。”","鲁宾森同时发出警示：人工智能能力的发展速度，已经超过研究人员对于对齐问题的认知水平；对齐这一研究领域致力于保证人工智能系统的行为符合人类的目标与价值观。"],"articleImages":[{"sourceUrl":"https://img.ithome.com/newsuploadfiles/2026/2/8ad2304d-c61b-4004-9483-e1a55ca43342.png?x-bce-process=image/format,f_auto","alt":"","afterParagraph":0,"url":"/media/articles/mqv2fnovlxtlcs8sgvw2mz47a/ff61fea60c25d5ed.webp"}],"mediaStatus":"ok","articleBodyZh":["前 OpenAI 安全员工 David Robinson 发文批评公司迭代部署过于激进、文化已崩坏 这条更新来自 ithome.com，发布时间为 2026-10-03，Aioga 保留原文入口以便核验。","摘要：据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 依赖先发布后加固的迭代部署模式，认为应像核电与航空业那样建立防护机制。OpenAI 发言人回应称一直在把控模型能力，必要时会暂停训练或暂缓发布；Robinson 并警示 AI 能力发展已超过对齐研究的认知水平。","背景：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。","Aioga 观察：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。","影响与后续：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。 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。"],"translationStatus":"","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 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":"2026-10-23T06:04:44.507Z","sourceHash":"d8a3b84b1cb3e24e","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["行业动态","IT之家"],"translations":{"zh-CN":{"title":"前 OpenAI 安全员工 David Robinson 发文批评公司迭代部署过于激进、文化已崩坏","summary":"据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 依赖先发布后加固的迭代部署模式，认为应像核电与航空业那样建立防护机制。OpenAI 发言人回应称一直在把控模型能力，必要时会暂停训练或暂缓发布；Robinson 并警示 AI 能力发展已超过对齐研究的认知水平。","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"前 OpenAI 安全员工 David Robinson 发文批评公司迭代部署过于激进、文化已崩坏 - Aioga AI资讯","description":"据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 依赖先发布后加固的迭代部署模式，认为应像核电与航空业那样建立防护机制。OpenAI 发言人回应称一直在把控模型能力，必要时会暂停训练或暂缓发布；Robinson 并警示 AI 能力发...","url":"https://www.aioga.com/news/mqv2fnovlxtlcs8sgvw2mz47a/","articleBody":["前 OpenAI 安全员工 David Robinson 发文批评公司迭代部署过于激进、文化已崩坏 这条更新来自 ithome.com，发布时间为 2026-10-03，Aioga 保留原文入口以便核验。","摘要：据路透社报道，在 OpenAI 任职三年半、曾监督 12 次前沿模型发布安全报告的前安全员工 David Robinson，当地时间周六在《大西洋》杂志发文，批评 OpenAI 依赖先发布后加固的迭代部署模式，认为应像核电与航空业那样建立防护机制。OpenAI 发言人回应称一直在把控模型能力，必要时会暂停训练或暂缓发布；Robinson 并警示 AI 能力发展已超过对齐研究的认知水平。","背景：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。","Aioga 观察：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。","影响与后续：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。 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。"]},"en":{"title":"Former OpenAI security employee David Robinson posted criticizing the company's iterative deployment as too aggressive and saying the culture has collapsed","summary":"According to Reuters, David Robinson, a former security employee who worked at OpenAI for three and a half years and oversaw 12 frontier model safety reports, published an article in The Atlantic magazine on Saturday local time, criticizing OpenAI for relying on an iterative deployment model of releasing first and then strengthening iterative deployment, arguing that protective mechanisms should be established like those in nuclear power and aviation. An OpenAI spokesperson responded that they have been monitoring model capabilities and will pause training or postpone releases if necessary; Robinson also warned that AI capability development has surpassed the cognitive level of alignment research.","category":"Industry","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Former OpenAI security employee David Robinson posted criticizing the company's iterative deployment as too aggressive and saying the culture has collapsed - Aioga AI News","description":"According to Reuters, David Robinson, a former security employee who worked at OpenAI for three and a half years and oversaw 12 frontier model safety reports, published an article...","url":"https://www.aioga.com/en/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:06.153Z"},"ja":{"title":"元OpenAIセキュリティ従業員のデイビッド・ロビンソン氏は、同社の反復展開を攻撃的すぎると批判し、文化が崩壊したと投稿しました","summary":"ロイターによると、OpenAIで3年半勤務し、12件のフロンティアモデル安全報告書を監督した元セキュリティ職員デイビッド・ロビンソン氏は、現地時間の土曜日にThe Atlantic誌に記事を掲載し、OpenAIがまずリリースし、その後強化する反復展開モデルに依存していることを批判し、原子力や航空のような保護メカニズムを確立すべきだと主張しました。OpenAIの広報担当者は、モデルの能力を監視しており、必要に応じてトレーニングを一時停止またはリリースを延期すると回答しました。また、AI能力開発はアライメント研究の認知レベルを超えていると警告しました。","category":"業界動向","source":"IT之家","aggregationSource":"IT之家","pageTitle":"元OpenAIセキュリティ従業員のデイビッド・ロビンソン氏は、同社の反復展開を攻撃的すぎると批判し、文化が崩壊したと投稿しました - Aioga AIニュース","description":"ロイターによると、OpenAIで3年半勤務し、12件のフロンティアモデル安全報告書を監督した元セキュリティ職員デイビッド・ロビンソン氏は、現地時間の土曜日にThe Atlantic誌に記事を掲載し、OpenAIがまずリリースし、その後強化する反復展開モデルに依存していることを批判し、原子力や航空のような保護メカニズムを確立すべきだと主張しました。OpenAI...","url":"https://www.aioga.com/ja/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:06.348Z"},"ko":{"title":"전 OpenAI 보안 직원 데이비드 로빈슨은 회사의 반복적 배포가 너무 공격적이라고 비판하며 문화가 무너졌다고 글을 올렸습니다","summary":"로이터에 따르면, OpenAI에서 3년 반 근무하며 12건의 프런티어 모델 안전 보고서를 감독한 전 보안 직원 데이비드 로빈슨은 현지 시간으로 토요일 The Atlantic 매거진에 기사를 게재해 OpenAI가 먼저 공개하고 그 다음에 반복 배포를 강화하는 반복적 배포 모델에 의존하는 것을 비판하며, 원자력 발전과 항공 분야처럼 보호 메커니즘을 구축해야 한다고 주장했다. OpenAI 대변인은 모델 역량을 모니터링해 왔으며 필요시 교육을 중단하거나 릴리스를 연기할 것이라고 답변했다; 로빈슨은 또한 AI 역량 개발이 정렬 연구의 인지 수준을 넘어섰다고 경고했다.","category":"업계 동향","source":"IT之家","aggregationSource":"IT之家","pageTitle":"전 OpenAI 보안 직원 데이비드 로빈슨은 회사의 반복적 배포가 너무 공격적이라고 비판하며 문화가 무너졌다고 글을 올렸습니다 - Aioga AI 뉴스","description":"로이터에 따르면, OpenAI에서 3년 반 근무하며 12건의 프런티어 모델 안전 보고서를 감독한 전 보안 직원 데이비드 로빈슨은 현지 시간으로 토요일 The Atlantic 매거진에 기사를 게재해 OpenAI가 먼저 공개하고 그 다음에 반복 배포를 강화하는 반복적 배포 모델에 의존하는 것을 비판하며, 원자력 발전과 항공...","url":"https://www.aioga.com/ko/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:06.039Z"},"es":{"title":"El exempleado de seguridad de OpenAI, David Robinson, publicó criticando el despliegue iterativo de la empresa por considerarlo demasiado agresivo y diciendo que la cultura se ha derrumbado","summary":"Según Reuters, David Robinson, un antiguo empleado de seguridad que trabajó en OpenAI durante tres años y medio y supervisó 12 informes de seguridad sobre modelos fronterizos, publicó un artículo en la revista The Atlantic el sábado hora local, criticando a OpenAI por confiar en un modelo de despliegue iterativo que primero libera y luego fortalece el despliegue iterativo, argumentando que deberían establecerse mecanismos de protección como los de la energía nuclear y la aviación. Un portavoz de OpenAI respondió que han estado monitorizando las capacidades del modelo y que pausarán el entrenamiento o pospondrán los lanzamientos si es necesario; Robinson también advirtió que el desarrollo de capacidades de IA ha superado el nivel cognitivo de la investigación sobre alineamiento.","category":"Industria","source":"IT之家","aggregationSource":"IT之家","pageTitle":"El exempleado de seguridad de OpenAI, David Robinson, publicó criticando el despliegue iterativo de la empresa por considerarlo demasiado agresivo y diciendo que la cultura se ha derrumbado - Aioga Noticias de IA","description":"Según Reuters, David Robinson, un antiguo empleado de seguridad que trabajó en OpenAI durante tres años y medio y supervisó 12 informes de seguridad sobre modelos fronterizos, publ...","url":"https://www.aioga.com/es/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:30.569Z"},"fr":{"title":"L’ancien employé de la sécurité d’OpenAI, David Robinson, a publié en critiquant le déploiement itératif de l’entreprise comme étant trop agressif et en affirmant que la culture s’était effondrée","summary":"Selon Reuters, David Robinson, ancien employé de la sécurité ayant travaillé chez OpenAI pendant trois ans et demi supervisé 12 rapports de sécurité sur les modèles frontières, a publié un article dans le magazine The Atlantic samedi, heure locale, critiquant OpenAI pour s’appuyer sur un modèle de déploiement itératif qui diffuse d’abord puis renforce le déploiement itératif, soutenant que des mécanismes de protection devraient être mis en place comme ceux de l’énergie nucléaire et de l’aviation. Un porte-parole d’OpenAI a répondu qu’ils surveillent les capacités du modèle et qu’ils mettraient en pause la formation ou reporteraient les sorties si nécessaire ; Robinson a également averti que le développement des capacités d’IA a dépassé le niveau cognitif de la recherche sur l’alignement.","category":"Industrie","source":"IT之家","aggregationSource":"IT之家","pageTitle":"L’ancien employé de la sécurité d’OpenAI, David Robinson, a publié en critiquant le déploiement itératif de l’entreprise comme étant trop agressif et en affirmant que la culture s’était effondrée - Aioga Actualités IA","description":"Selon Reuters, David Robinson, ancien employé de la sécurité ayant travaillé chez OpenAI pendant trois ans et demi supervisé 12 rapports de sécurité sur les modèles frontières, a p...","url":"https://www.aioga.com/fr/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:30.172Z"},"de":{"title":"Der ehemalige OpenAI-Sicherheitsmitarbeiter David Robinson postete, in dem er die iterative Einführung des Unternehmens als zu aggressiv kritisierte und sagte, die Unternehmenskultur sei zusammengebrochen","summary":"Laut Reuters veröffentlichte David Robinson, ein ehemaliger Sicherheitsmitarbeiter, der dreieinhalb Jahre bei OpenAI arbeitete und zwölf Sicherheitsberichte für Frontier-Modelle beaufsichtigte, am Samstag einen Artikel im Atlantic-Magazin Ortszeit, in dem er OpenAI dafür kritisierte, auf ein iteratives Deployment-Modell zu setzen, bei dem die iterative Deployment zuerst freigegeben und dann die iterative Deployment verstärkt wird, und argumentierte, dass Schutzmechanismen wie in der Kernkraft und Luftfahrt etabliert werden sollten. Ein OpenAI-Sprecher antwortete, dass sie die Modellfähigkeiten überwachen und das Training pausieren oder die Veröffentlichungen bei Bedarf verschieben werden; Robinson warnte außerdem, dass die Entwicklung von KI-Fähigkeiten das kognitive Niveau der Ausrichtungsforschung überschritten habe.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Der ehemalige OpenAI-Sicherheitsmitarbeiter David Robinson postete, in dem er die iterative Einführung des Unternehmens als zu aggressiv kritisierte und sagte, die Unternehmenskultur sei zusammengebrochen - Aioga KI-News","description":"Laut Reuters veröffentlichte David Robinson, ein ehemaliger Sicherheitsmitarbeiter, der dreieinhalb Jahre bei OpenAI arbeitete und zwölf Sicherheitsberichte für Frontier-Modelle be...","url":"https://www.aioga.com/de/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:30.915Z"},"pt-BR":{"title":"O ex-funcionário de segurança da OpenAI, David Robinson, publicou criticando a implantação iterativa da empresa por ser agressiva demais e dizendo que a cultura entrou em colapso","summary":"Segundo a Reuters, David Robinson, ex-funcionário de segurança que trabalhou na OpenAI por três anos e meio e supervisionou 12 relatórios de segurança de modelos de fronteira, publicou um artigo na revista The Atlantic no sábado, horário local, criticando a OpenAI por depender de um modelo de implantação iterativa que consiste em lançar primeiro e depois fortalecer a implantação iterativa, argumentando que mecanismos de proteção devem ser estabelecidos como os da energia nuclear e da aviação. Um porta-voz da OpenAI respondeu que eles têm monitorado as capacidades do modelo e que podem pausar o treinamento ou adiar os lançamentos se necessário; Robinson também alertou que o desenvolvimento de capacidades de IA ultrapassou o nível cognitivo da pesquisa em alinhamento.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"O ex-funcionário de segurança da OpenAI, David Robinson, publicou criticando a implantação iterativa da empresa por ser agressiva demais e dizendo que a cultura entrou em colapso - Aioga Notícias de IA","description":"Segundo a Reuters, David Robinson, ex-funcionário de segurança que trabalhou na OpenAI por três anos e meio e supervisionou 12 relatórios de segurança de modelos de fronteira, publ...","url":"https://www.aioga.com/pt-BR/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:54.428Z"},"ru":{"title":"Бывший сотрудник по безопасности OpenAI Дэвид Робинсон раскритиковал итеративное внедрение компании как слишком агрессивное и отметил, что культура рухнула","summary":"По данным Reuters, Дэвид Робинсон, бывший сотрудник службы безопасности в OpenAI, проработавший в OpenAI три с половиной года и курировавший 12 отчётов по безопасности на фронтирных моделях, опубликовал статью в журнале The Atlantic в субботу по местному времени, критикуя OpenAI за опору на итеративную модель развертывания — сначала выпускать и затем усиливать итеративное развертывание, утверждая, что необходимо создать защитные механизмы, подобные тем, что используются в ядерной энергетике и авиации. Представитель OpenAI ответил, что они отслеживают возможности модели и при необходимости приостанут обучение или отложат выпуски; Робинсон также предупредил, что развитие возможностей ИИ превысило когнитивный уровень исследований согласования.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Бывший сотрудник по безопасности OpenAI Дэвид Робинсон раскритиковал итеративное внедрение компании как слишком агрессивное и отметил, что культура рухнула - Aioga Новости ИИ","description":"По данным Reuters, Дэвид Робинсон, бывший сотрудник службы безопасности в OpenAI, проработавший в OpenAI три с половиной года и курировавший 12 отчётов по безопасности на фронтирны...","url":"https://www.aioga.com/ru/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:54.645Z"},"ar":{"title":"نشر ديفيد روبنسون، موظف الأمن السابق في OpenAI، انتقد النشر التكراري للشركة واعتبرته عدوانيا جدا قائلا إن الثقافة انهارت","summary":"وفقا لرويترز، نشر ديفيد روبنسون، موظف أمني سابق عمل في OpenAI لمدة ثلاث سنوات ونصف وأشرف على 12 تقريرا لسلامة النماذج الحدودية، مقالا في مجلة The Atlantic يوم السبت بالتوقيت المحلي، ينتقد OpenAI لاعتماده على نموذج نشر تكراري يطلق أولا ثم تقوية النشر التكراري، بحجة أنه يجب إنشاء آليات حماية مثل تلك في الطاقة النووية والطيران. رد متحدث باسم OpenAI بأنهم يراقبون قدرات النماذج وسيؤجلون التدريب مؤقتا أو يؤجلون الإطلاقات إذا لزم الأمر؛ حذر روبنسون أيضا من أن تطوير قدرات الذكاء الاصطناعي تجاوز المستوى المعرفي لأبحاث المحاذاة.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"نشر ديفيد روبنسون، موظف الأمن السابق في OpenAI، انتقد النشر التكراري للشركة واعتبرته عدوانيا جدا قائلا إن الثقافة انهارت - Aioga أخبار الذكاء الاصطناعي","description":"وفقا لرويترز، نشر ديفيد روبنسون، موظف أمني سابق عمل في OpenAI لمدة ثلاث سنوات ونصف وأشرف على 12 تقريرا لسلامة النماذج الحدودية، مقالا في مجلة The Atlantic يوم السبت بالتوقيت المحلي...","url":"https://www.aioga.com/ar/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:01:55.047Z"},"hi":{"title":"OpenAI के पूर्व सुरक्षा कर्मचारी डेविड रॉबिन्सन ने कंपनी की पुनरावृत्त तैनाती की आलोचना करते हुए पोस्ट किया और कहा कि संस्कृति ध्वस्त हो गई है","summary":"रॉयटर्स के अनुसार, डेविड रॉबिन्सन, एक पूर्व सुरक्षा कर्मचारी, जिन्होंने साढ़े तीन साल तक ओपनएआई में काम किया और 12 फ्रंटियर मॉडल सुरक्षा रिपोर्टों की देखरेख की, ने शनिवार को स्थानीय समय पर द अटलांटिक पत्रिका में एक लेख प्रकाशित किया, जिसमें पहले जारी करने और फिर पुनरावृत्त तैनाती को मजबूत करने के पुनरावृत्त परिनियोजन मॉडल पर भरोसा करने के लिए ओपनएआई की आलोचना की गई, यह तर्क देते हुए कि परमाणु ऊर्जा और विमानन की तरह सुरक्षात्मक तंत्र स्थापित किया जाना चाहिए। एक ओपनएआई प्रवक्ता ने जवाब दिया कि वे मॉडल क्षमताओं की निगरानी कर रहे हैं और यदि आवश्यक हो तो प्रशिक्षण को रोक देंगे या रिलीज को स्थगित कर देंगे; रॉबिन्सन ने यह भी चेतावनी दी कि एआई क्षमता विकास संरेखण अनुसंधान के संज्ञानात्मक स्तर को पार कर गया है।","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"OpenAI के पूर्व सुरक्षा कर्मचारी डेविड रॉबिन्सन ने कंपनी की पुनरावृत्त तैनाती की आलोचना करते हुए पोस्ट किया और कहा कि संस्कृति ध्वस्त हो गई है - Aioga AI समाचार","description":"रॉयटर्स के अनुसार, डेविड रॉबिन्सन, एक पूर्व सुरक्षा कर्मचारी, जिन्होंने साढ़े तीन साल तक ओपनएआई में काम किया और 12 फ्रंटियर मॉडल सुरक्षा रिपोर्टों की देखरेख की, ने शनिवार को स्थानी...","url":"https://www.aioga.com/hi/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:02:19.064Z"},"it":{"title":"L'ex dipendente della sicurezza di OpenAI David Robinson ha pubblicato criticando il deployment iterativo dell'azienda definendolo troppo aggressivo e affermando che la cultura è crollata","summary":"Secondo Reuters, David Robinson, ex dipendente della sicurezza che ha lavorato per tre anni e mezzo in OpenAI e ha supervisionato 12 rapporti sulla sicurezza dei modelli di frontiera, ha pubblicato un articolo sulla rivista The Atlantic sabato ora locale, criticando OpenAI per aver fatto affidamento su un modello di deployment iterativo che prevede il rilascio prima e poi il rafforzamento del deployment iterativo, sostenendo che dovrebbero essere stabiliti meccanismi di protezione come quelli dell'energia nucleare e dell'aviazione. Un portavoce di OpenAI ha risposto che stanno monitorando le capacità del modello e che sospenderanno l'addestramento o posticiperanno le release se necessario; Robinson ha anche avvertito che lo sviluppo delle capacità di IA ha superato il livello cognitivo della ricerca sull'allineamento.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"L'ex dipendente della sicurezza di OpenAI David Robinson ha pubblicato criticando il deployment iterativo dell'azienda definendolo troppo aggressivo e affermando che la cultura è crollata - Aioga Notizie IA","description":"Secondo Reuters, David Robinson, ex dipendente della sicurezza che ha lavorato per tre anni e mezzo in OpenAI e ha supervisionato 12 rapporti sulla sicurezza dei modelli di frontie...","url":"https://www.aioga.com/it/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:02:19.013Z"},"nl":{"title":"Voormalig OpenAI-beveiligingsmedewerker David Robinson plaatste kritiek op de iteratieve implementatie van het bedrijf als te agressief en zei dat de cultuur is ingestort","summary":"Volgens Reuters publiceerde David Robinson, een voormalig beveiligingsmedewerker die drieënhalf jaar bij OpenAI werkte en toezicht hield op 12 veiligheidsrapporten over frontiermodellen, zaterdag lokale tijd een artikel in The Atlantic magazine, waarin hij OpenAI bekritiseerde omdat het vertrouwt op een iteratief implementatiemodel waarbij eerst wordt vrijgegeven en daarna de iteratieve implementatie wordt versterkt, en betoogde dat beschermende mechanismen moeten worden ingesteld zoals die in kernenergie en luchtvaart. Een woordvoerder van OpenAI reageerde dat ze de modelmogelijkheden monitoren en de training zullen pauzeren of releases zullen uitstellen indien nodig; Robinson waarschuwde ook dat de ontwikkeling van AI-capaciteiten het cognitieve niveau van alignementonderzoek heeft overtroffen.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Voormalig OpenAI-beveiligingsmedewerker David Robinson plaatste kritiek op de iteratieve implementatie van het bedrijf als te agressief en zei dat de cultuur is ingestort - Aioga AI-nieuws","description":"Volgens Reuters publiceerde David Robinson, een voormalig beveiligingsmedewerker die drieënhalf jaar bij OpenAI werkte en toezicht hield op 12 veiligheidsrapporten over frontiermod...","url":"https://www.aioga.com/nl/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:02:19.153Z"},"tr":{"title":"Eski OpenAI güvenlik çalışanı David Robinson, şirketin yinelemeli dağıtımını fazla agresif olmakla eleştirerek kültürün çöktüğünü söyledi","summary":"Reuters'a göre, OpenAI'de üç buçuk yıl çalışmış ve 12 sınır model güvenlik raporunu denetleyen eski güvenlik çalışanı David Robinson, yerel saatle Cumartesi günü The Atlantic dergisinde bir makale yayımladı; OpenAI'yi, önce serbest bırakıp sonra güçlendiren yinelemeli bir dağıtım modeline dayandığı için eleştirdi ve nükleer enerji ve havacılıkta olduğu gibi koruyucu mekanizmaların kurulması gerektiğini savundu. OpenAI sözcüsü, model yeteneklerini izlediklerini ve gerekirse eğitimi durduracaklarını veya yayınları erteleyeceklerini söyledi; Robinson ayrıca yapay zeka yetenek geliştirmenin uyum araştırmalarının bilişsel seviyesini aştığını uyardı.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Eski OpenAI güvenlik çalışanı David Robinson, şirketin yinelemeli dağıtımını fazla agresif olmakla eleştirerek kültürün çöktüğünü söyledi - Aioga AI Haberleri","description":"Reuters'a göre, OpenAI'de üç buçuk yıl çalışmış ve 12 sınır model güvenlik raporunu denetleyen eski güvenlik çalışanı David Robinson, yerel saatle Cumartesi günü The Atlantic dergi...","url":"https://www.aioga.com/tr/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:02:43.203Z"},"vi":{"title":"Cựu nhân viên bảo mật OpenAI, David Robinson, đã đăng bài chỉ trích việc triển khai lặp đi lặp lại của công ty là quá hung hăng và cho rằng văn hóa đã sụp đổ","summary":"Theo Reuters, David Robinson, một cựu nhân viên an ninh làm việc tại OpenAI trong ba năm rưỡi và giám sát 12 báo cáo an toàn mô hình biên giới, đã đăng một bài báo trên tạp chí The Atlantic vào thứ Bảy theo giờ địa phương, chỉ trích OpenAI vì dựa vào mô hình triển khai lặp đi lặp lại, phát hành trước rồi mới tăng cường triển khai lặp lại, lập luận rằng nên thiết lập các cơ chế bảo vệ giống như trong năng lượng hạt nhân và hàng không. Người phát ngôn của OpenAI trả lời rằng họ đã giám sát khả năng mô hình và sẽ tạm dừng đào tạo hoặc hoãn phát hành nếu cần thiết; Robinson cũng cảnh báo rằng phát triển năng lực AI đã vượt qua mức độ nhận thức của nghiên cứu căn chỉnh.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Cựu nhân viên bảo mật OpenAI, David Robinson, đã đăng bài chỉ trích việc triển khai lặp đi lặp lại của công ty là quá hung hăng và cho rằng văn hóa đã sụp đổ - Tin tức AI Aioga","description":"Theo Reuters, David Robinson, một cựu nhân viên an ninh làm việc tại OpenAI trong ba năm rưỡi và giám sát 12 báo cáo an toàn mô hình biên giới, đã đăng một bài báo trên tạp chí The...","url":"https://www.aioga.com/vi/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:02:43.456Z"},"id":{"title":"Mantan karyawan keamanan OpenAI, David Robinson, mengkritik penerapan iteratif perusahaan sebagai terlalu agresif dan mengatakan budaya tersebut telah runtuh","summary":"Menurut Reuters, David Robinson, mantan pegawai keamanan yang bekerja di OpenAI selama tiga setengah tahun dan mengawasi 12 laporan keselamatan model perbatasan, menerbitkan artikel di majalah The Atlantic pada hari Sabtu waktu setempat, mengkritik OpenAI karena mengandalkan model penerapan iteratif yang merilis terlebih dahulu lalu memperkuat penerapan iteratif, dengan berargumen bahwa mekanisme perlindungan harus dibangun seperti di tenaga nuklir dan penerbangan. Juru bicara OpenAI menanggapi bahwa mereka telah memantau kemampuan model dan akan menghentikan pelatihan atau menunda rilis jika diperlukan; Robinson juga memperingatkan bahwa pengembangan kemampuan AI telah melampaui tingkat kognitif penelitian penyelarasan.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Mantan karyawan keamanan OpenAI, David Robinson, mengkritik penerapan iteratif perusahaan sebagai terlalu agresif dan mengatakan budaya tersebut telah runtuh - Berita AI Aioga","description":"Menurut Reuters, David Robinson, mantan pegawai keamanan yang bekerja di OpenAI selama tiga setengah tahun dan mengawasi 12 laporan keselamatan model perbatasan, menerbitkan artike...","url":"https://www.aioga.com/id/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:02:43.378Z"},"th":{"title":"อดีตพนักงานด้านความปลอดภัยของ OpenAI เดวิด โรบินสัน โพสต์วิจารณ์การปรับใช้แบบซ้ํา ๆ ของบริษัทว่าก้าวร้าวเกินไป และกล่าวว่าวัฒนธรรมองค์กรได้ล่มสลายแล้ว","summary":"ตามรายงานของรอยเตอร์ เดวิด โรบินสัน อดีตพนักงานด้านความปลอดภัยที่ทํางานที่ OpenAI เป็นเวลาสามปีครึ่งและดูแลรายงานความปลอดภัยของโมเดลแนวหน้า 12 ฉบับ ได้ตีพิมพ์บทความในนิตยสาร The Atlantic เมื่อวันเสาร์ตามเวลาท้องถิ่น วิจารณ์ OpenAI ที่พึ่งพาโมเดลการปรับใช้แบบวนซ้ํา โดยปล่อยก่อนแล้วจึงเสริมความแข็งแกร่งในการปรับใช้แบบวนซ้ํา โดยโต้แย้งว่าควรมีการสร้างกลไกป้องกันเหมือนกับในพลังงานนิวเคลียร์และการบินโฆษกของ OpenAI ตอบว่าพวกเขากําลังติดตามความสามารถของโมเดลและจะหยุดการฝึกอบรมหรือเลื่อนการปล่อยหากจําเป็น;โรบินสันยังเตือนว่าการพัฒนาความสามารถของ AI ได้ก้าวข้ามระดับความรู้ความเข้าใจของการวิจัยการจัดแนวแล้ว","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"อดีตพนักงานด้านความปลอดภัยของ OpenAI เดวิด โรบินสัน โพสต์วิจารณ์การปรับใช้แบบซ้ํา ๆ ของบริษัทว่าก้าวร้าวเกินไป และกล่าวว่าวัฒนธรรมองค์กรได้ล่มสลายแล้ว - ข่าว AI Aioga","description":"ตามรายงานของรอยเตอร์ เดวิด โรบินสัน อดีตพนักงานด้านความปลอดภัยที่ทํางานที่ OpenAI เป็นเวลาสามปีครึ่งและดูแลรายงานความปลอดภัยของโมเดลแนวหน้า 12 ฉบับ ได้ตีพิมพ์บทความในนิตยสาร The At...","url":"https://www.aioga.com/th/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:03:07.271Z"},"pl":{"title":"Były pracownik ds. bezpieczeństwa OpenAI, David Robinson, opublikował wpis, w którym krytykował iteracyjne wdrożenie firmy jako zbyt agresywne i twierdził, że kultura upadła","summary":"Według Reutersa, David Robinson, były pracownik ds. bezpieczeństwa, który pracował w OpenAI przez trzy i pół roku i nadzorował 12 raportów dotyczących bezpieczeństwa modeli granicznych, opublikował artykuł w magazynie The Atlantic w sobotę czasu lokalnego, krytykując OpenAI za poleganie na modelu wdrożenia iteracyjnego, polegającym na jego wypuszczaniu najpierw, a następnie wzmacnianiu, argumentując, że należy wprowadzać mechanizmy ochronne podobne do tych w energetyce jądrowej i lotnictwie. Rzecznik OpenAI odpowiedział, że monitorują możliwości modeli i w razie potrzeby wstrzymają szkolenia lub odroczą publikacje; Robinson ostrzegł również, że rozwój zdolności AI przewyższa poziom poznawczy w badaniach nad dopasowaniem.","category":"行业动态","source":"IT之家","aggregationSource":"IT之家","pageTitle":"Były pracownik ds. bezpieczeństwa OpenAI, David Robinson, opublikował wpis, w którym krytykował iteracyjne wdrożenie firmy jako zbyt agresywne i twierdził, że kultura upadła - Aioga Wiadomości AI","description":"Według Reutersa, David Robinson, były pracownik ds. bezpieczeństwa, który pracował w OpenAI przez trzy i pół roku i nadzorował 12 raportów dotyczących bezpieczeństwa modeli granicz...","url":"https://www.aioga.com/pl/news/mqv2fnovlxtlcs8sgvw2mz47a/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"cee828378a792d7d","translatedAt":"2026-10-04T03:03:07.436Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}