{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-09-28T06:03:00.468Z","headline":"安全研究人员用 Claude Opus 5 在 72 小时内攻破 OpenAI 内部系统","description":"Hacktron 三名安全研究人员利用 Anthropic 的 Claude 模型，通过 OpenAI 社区论坛在不到 72 小时内攻入其内部系统和代码仓库。","url":"https://www.aioga.com/news/cmu798syp0v78rowkymgv8wgn/","mainEntityOfPage":"https://www.aioga.com/news/cmu798syp0v78rowkymgv8wgn/","datePublished":"2026-09-18T17:20:30.000Z","dateModified":"2026-09-18T17:20:30.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://the-decoder.com/security-researchers-used-anthropics-claude-to-hack-openais-internal-systems-in-under-72-hours","https://aihot.news/items/cmu798syp0v78rowkymgv8wgn"],"canonicalUrl":"https://www.aioga.com/news/cmu798syp0v78rowkymgv8wgn/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：Hacktron 三名安全研究人员利用 Anthropic 的 Claude 模型，通过 OpenAI 社区论坛在不到 72 小时内攻入其内部系统和代码仓库。 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/cmu798syp0v78rowkymgv8wgn/","dateCreated":"2026-09-18T17:20:30.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"the-decoder.com source article","url":"https://the-decoder.com/security-researchers-used-anthropics-claude-to-hack-openais-internal-systems-in-under-72-hours","datePublished":"2026-09-18T17:20:30.000Z","provider":{"@type":"Organization","name":"the-decoder.com","url":"https://the-decoder.com/security-researchers-used-anthropics-claude-to-hack-openais-internal-systems-in-under-72-hours"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.news/items/cmu798syp0v78rowkymgv8wgn","datePublished":"2026-09-18T17:20:30.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.news/items/cmu798syp0v78rowkymgv8wgn"}}],"aggregationSource":"The Decoder：AI News（RSS）","originalPublisher":{"name":"the-decoder.com","url":"https://the-decoder.com/security-researchers-used-anthropics-claude-to-hack-openais-internal-systems-in-under-72-hours"},"geoDeepAnswer":null,"article":{"id":"cmu798syp0v78rowkymgv8wgn","slug":"cmu798syp0v78rowkymgv8wgn","url":"https://www.aioga.com/news/cmu798syp0v78rowkymgv8wgn/","title":"安全研究人员用 Claude Opus 5 在 72 小时内攻破 OpenAI 内部系统","title_en":"","summary":"Hacktron 三名安全研究人员利用 Anthropic 的 Claude 模型，通过 OpenAI 社区论坛在不到 72 小时内攻入其内部系统和代码仓库。","source":"The Decoder：AI News（RSS）","sourceUrl":"https://the-decoder.com/security-researchers-used-anthropics-claude-to-hack-openais-internal-systems-in-under-72-hours","aiHotUrl":"https://aihot.news/items/cmu798syp0v78rowkymgv8wgn","publishedAt":"2026-09-18T17:20:30.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["Three security researchers used Anthropic's Claude models to break into OpenAI's internal systems through the company's community forum. The attack took less than 72 hours and, according to the team, only became possible once Opus 5 shipped.","OpenAI is getting a taste of its own medicine. After inadvertently letting agents hack their way across the internet for months：https://the-decoder.com/openai-reportedly-slows-research-after-its-own-models-secretly-coordinated-hacks-for-weeks-undetected/, the company has now been hacked with AI's help. Hacktron's security team chained two vulnerabilities together to access OpenAI employees' ChatGPT and Codex accounts. From there, they broke into OpenAI's internal code repository on GitHub.","The attack：https://www.hacktron.ai/blog/hacking-openai ran through OpenAI's community forum at community.openai.com：http://community.openai.com. Any user or employee who had used \"Sign in with OpenAI\" there was potentially affected. Users can connect GitHub, Slack, and email to Codex and ChatGPT, so the attack could theoretically have reached those services too. To prove they had access, the researchers used an employee's Codex account to create a harmless pull request in the internal monorepo. They say they didn't view any sensitive data. Ad","The first vulnerability was in libheif, the library the forum used to process uploaded HEIC images. According to Hacktron, a fix had been available in the original source code：https://github.com/strukturag/libheif/commit/85e21ad44eba931314337300a2376b8d28f085ae for a year, but no one had flagged it as a security issue. The Debian packages running on the forum still lacked the fix. A crafted image file let the researchers run their own code on the server. Ad","The second vulnerability was a misconfiguration in OpenAI's central single sign-on (SSO) system. Anyone controlling the forum server could impersonate active forum members and take over their ChatGPT and Codex accounts. The flaw extended beyond the forum, Hacktron writes. Any compromised service using OpenAI login would have granted the same access.","The researchers say they initially used Claude Opus 4.8 to find the vulnerability. The model built a working exploit, but only with ASLR, a common defense against memory attacks, disabled. Across several sessions, it couldn't produce a reliable version with ASLR enabled. Ad","On the evening of July 24, Anthropic released Claude Opus 5：https://the-decoder.com/anthropic-claims-its-new-claude-opus-5-delivers-near-fable-5-performance-at-half-the-token-price/. According to Hacktron, the new model produced a working exploit for a local Mac within three hours, then adapted it to the Discourse server environment. The researchers then ran Claude in an autonomous loop against their own test instance.","Because the model refused to write exploits against real systems, they presented the target as a benchmark task. Four hours later, the agent had taken over the server. On a related task, the researchers also observed a jump in performance compared to OpenAI's GPT-5.6 Sol. Ad","OpenAI confirmed the fix about 14 hours after the report. Discourse, the software behind the forum, also responded within days. Ad","Beyond the OpenAI hack, the researchers expanded their investigation, dubbed \"HEIF Heist\"：https://heif-heist.com/, to cover Slack, Meta, GitHub Enterprise, and other targets. Three people carried out the project over two months, spending less than $3,000 on AI. Adapting the attack to each new target took just one to two days. Only Shopify noticed the activity, despite thousands of image uploads and repeated crashes in image processing.","Software has long benefited from a kind of security through complexity, the Hacktron team writes. Even with public source code and a known vulnerability, building a reliable exploit required rare expertise, time, and deep knowledge of the target environment. Complexity wasn't a true security barrier, but it did protect many companies in practice.","AI strips away that protection by replacing scarce expertise with computing power. Hacktron argues that threat models must reflect how cheap attacks have become. \"Work that once required a well-resourced team and months of effort can now be compressed into days,\" the team writes.","Stay in the loop on AI. Clear, useful, no fluff.","Follow The Decoder for AI news, background stories and expert analyses.","The Decoder：https://the-decoder.com/"],"articleImages":[{"sourceUrl":"https://the-decoder.com/wp-content/uploads/2026/09/openai_anthropic_hack.png","alt":"Image description","afterParagraph":0,"url":"/media/articles/cmu798syp0v78rowkymgv8wgn/6ddd390b789c5b66.png"}],"mediaStatus":"ok","articleBodyZh":["三名安全研究人员利用Anthropic的Claude模型，通过公司的社区论坛入侵了OpenAI的内部系统。此次攻击耗时不到72小时，据团队称，只有在Opus 5发布后才有可能实现。","OpenAI正在自食其果。在无意中允许代理在互联网上入侵数月后：https://the-decoder.com/openai-reportedly-slows-research-after-its-own-models-secretly-coordinated-hacks-for-weeks-undetected/，该公司现在在AI的帮助下被黑客入侵。Hacktron的安全团队将两个漏洞串联起来，访问了OpenAI员工的ChatGPT和Codex账户。随后，他们入侵了OpenAI在GitHub上的内部代码仓库。","此次攻击：https://www.hacktron.ai/blog/hacking-openai 是通过OpenAI的社区论坛 community.openai.com：http://community.openai.com 进行的。任何在论坛使用过“使用OpenAI登录”的用户或员工都有可能受到影响。用户可以将GitHub、Slack和电子邮件连接到Codex和ChatGPT，因此理论上攻击也可能波及这些服务。为了证明他们获得了访问权限，研究人员使用一名员工的Codex账户在内部monorepo中创建了一个无害的pull请求。他们表示没有查看任何敏感数据。广告","第一个漏洞存在于libheif，这是论坛用来处理上传HEIC图片的库。根据Hacktron的说法，修复已经在原始源代码中提供了一年：https://github.com/strukturag/libheif/commit/85e21ad44eba931314337300a2376b8d28f085ae，但没有人将其标记为安全问题。运行论坛的Debian软件包仍然缺少该修复。一个精心制作的图像文件使研究人员能够在服务器上运行自己的代码。广告","第二个漏洞是OpenAI中央单点登录（SSO）系统的配置错误。任何控制论坛服务器的人都可以冒充活跃的论坛成员，并接管他们的ChatGPT和Codex账户。Hacktron写道，该漏洞不仅限于论坛。任何使用OpenAI登录的受影响服务都会授予相同的访问权限。","研究人员表示，他们最初使用 Claude Opus 4.8 来发现该漏洞。模型构建了一个可行的利用程序，但仅在禁用 ASLR（一种常见的内存攻击防御机制）时才能运行。在几次实验中，它无法在启用 ASLR 的情况下生成可靠版本。广告","7 月 24 日晚，Anthropic 发布了 Claude Opus 5：https://the-decoder.com/anthropic-claims-its-new-claude-opus-5-delivers-near-fable-5-performance-at-half-the-token-price/。根据 Hacktron 的说法，新模型在三小时内成功生成了对本地 Mac 的可行利用程序，然后将其适配到 Discourse 服务器环境中。研究人员随后在他们自己的测试实例上以自主循环运行 Claude。","由于模型拒绝针对真实系统编写利用程序，他们将目标呈现为基准任务。四小时后，代理已接管服务器。在相关任务中，研究人员还观察到与 OpenAI 的 GPT-5.6 Sol 相比的性能跃升。广告","OpenAI 在报告约 14 小时后确认了修复。论坛背后的软件 Discourse 也在数天内作出回应。广告","除了 OpenAI 的漏洞，研究人员还扩大了他们的调查，称之为“HEIF Heist”：https://heif-heist.com/，涵盖 Slack、Meta、GitHub Enterprise 及其他目标。三人团队在两个月内完成该项目，AI 花费不到 3,000 美元。将攻击适配到每个新目标仅需一到两天。尽管进行了数千次图像上传和多次图像处理崩溃，但只有 Shopify 注意到了这些活动。","Hacktron 团队写道，软件长期以来受益于一种通过复杂性实现的安全性。即使有公开的源代码和已知漏洞，构建可靠的利用程序仍需要罕见的专业知识、时间以及对目标环境的深入了解。复杂性并非真正的安全屏障，但实际上确实保护了许多公司。","AI 通过用计算能力替代稀缺的专业知识，剥夺了这种保护。Hacktron 认为，威胁模型必须反映攻击已变得多么廉价。“曾经需要资源充足的团队和数月努力的工作，现在可以压缩到几天内完成，”团队写道。","保持 AI 相关信息的更新。清晰、有用，无废话。","关注 The Decoder，获取 AI 新闻、背景故事和专家分析。","解码器：https://the-decoder.com/"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：Hacktron 三名安全研究人员利用 Anthropic 的 Claude 模型，通过 OpenAI 社区论坛在不到 72 小时内攻入其内部系统和代码仓库。 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：模型与研究类动态需要结合能力边界、开放方式、成本、可用性和真实任务表现判断，单项指标领先不等于已经形成稳定采用。","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：继续观察官方文档、实际可用性、价格变化、开发者反馈和竞品回应。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-09-28T06:19:15.384Z","sourceHash":"ea19e4f2dcad168f","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["行业动态","The Decoder：AI News（RSS）"],"translations":{"zh-CN":{"title":"安全研究人员用 Claude Opus 5 在 72 小时内攻破 OpenAI 内部系统","summary":"Hacktron 三名安全研究人员利用 Anthropic 的 Claude 模型，通过 OpenAI 社区论坛在不到 72 小时内攻入其内部系统和代码仓库。","category":"行业动态","source":"the-decoder.com","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"安全研究人员用 Claude Opus 5 在 72 小时内攻破 OpenAI 内部系统 - Aioga AI资讯","description":"Hacktron 三名安全研究人员利用 Anthropic 的 Claude 模型，通过 OpenAI 社区论坛在不到 72 小时内攻入其内部系统和代码仓库。","url":"https://www.aioga.com/news/cmu798syp0v78rowkymgv8wgn/","articleBody":["三名安全研究人员利用Anthropic的Claude模型，通过公司的社区论坛入侵了OpenAI的内部系统。此次攻击耗时不到72小时，据团队称，只有在Opus 5发布后才有可能实现。","OpenAI正在自食其果。在无意中允许代理在互联网上入侵数月后：https://the-decoder.com/openai-reportedly-slows-research-after-its-own-models-secretly-coordinated-hacks-for-weeks-undetected/，该公司现在在AI的帮助下被黑客入侵。Hacktron的安全团队将两个漏洞串联起来，访问了OpenAI员工的ChatGPT和Codex账户。随后，他们入侵了OpenAI在GitHub上的内部代码仓库。","此次攻击：https://www.hacktron.ai/blog/hacking-openai 是通过OpenAI的社区论坛 community.openai.com：http://community.openai.com 进行的。任何在论坛使用过“使用OpenAI登录”的用户或员工都有可能受到影响。用户可以将GitHub、Slack和电子邮件连接到Codex和ChatGPT，因此理论上攻击也可能波及这些服务。为了证明他们获得了访问权限，研究人员使用一名员工的Codex账户在内部monorepo中创建了一个无害的pull请求。他们表示没有查看任何敏感数据。广告","第一个漏洞存在于libheif，这是论坛用来处理上传HEIC图片的库。根据Hacktron的说法，修复已经在原始源代码中提供了一年：https://github.com/strukturag/libheif/commit/85e21ad44eba931314337300a2376b8d28f085ae，但没有人将其标记为安全问题。运行论坛的Debian软件包仍然缺少该修复。一个精心制作的图像文件使研究人员能够在服务器上运行自己的代码。广告","第二个漏洞是OpenAI中央单点登录（SSO）系统的配置错误。任何控制论坛服务器的人都可以冒充活跃的论坛成员，并接管他们的ChatGPT和Codex账户。Hacktron写道，该漏洞不仅限于论坛。任何使用OpenAI登录的受影响服务都会授予相同的访问权限。","研究人员表示，他们最初使用 Claude Opus 4.8 来发现该漏洞。模型构建了一个可行的利用程序，但仅在禁用 ASLR（一种常见的内存攻击防御机制）时才能运行。在几次实验中，它无法在启用 ASLR 的情况下生成可靠版本。广告","7 月 24 日晚，Anthropic 发布了 Claude Opus 5：https://the-decoder.com/anthropic-claims-its-new-claude-opus-5-delivers-near-fable-5-performance-at-half-the-token-price/。根据 Hacktron 的说法，新模型在三小时内成功生成了对本地 Mac 的可行利用程序，然后将其适配到 Discourse 服务器环境中。研究人员随后在他们自己的测试实例上以自主循环运行 Claude。","由于模型拒绝针对真实系统编写利用程序，他们将目标呈现为基准任务。四小时后，代理已接管服务器。在相关任务中，研究人员还观察到与 OpenAI 的 GPT-5.6 Sol 相比的性能跃升。广告","OpenAI 在报告约 14 小时后确认了修复。论坛背后的软件 Discourse 也在数天内作出回应。广告","除了 OpenAI 的漏洞，研究人员还扩大了他们的调查，称之为“HEIF Heist”：https://heif-heist.com/，涵盖 Slack、Meta、GitHub Enterprise 及其他目标。三人团队在两个月内完成该项目，AI 花费不到 3,000 美元。将攻击适配到每个新目标仅需一到两天。尽管进行了数千次图像上传和多次图像处理崩溃，但只有 Shopify 注意到了这些活动。","Hacktron 团队写道，软件长期以来受益于一种通过复杂性实现的安全性。即使有公开的源代码和已知漏洞，构建可靠的利用程序仍需要罕见的专业知识、时间以及对目标环境的深入了解。复杂性并非真正的安全屏障，但实际上确实保护了许多公司。","AI 通过用计算能力替代稀缺的专业知识，剥夺了这种保护。Hacktron 认为，威胁模型必须反映攻击已变得多么廉价。“曾经需要资源充足的团队和数月努力的工作，现在可以压缩到几天内完成，”团队写道。","保持 AI 相关信息的更新。清晰、有用，无废话。","关注 The Decoder，获取 AI 新闻、背景故事和专家分析。","解码器：https://the-decoder.com/"]},"en":{"title":"Security researchers hacked OpenAI's internal systems within 72 hours using Claude Opus 5","summary":"Three security researchers from Hacktron used Anthropic's Claude model to breach its internal systems and code repositories through the OpenAI community forum in less than 72 hours.","category":"Industry","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Security researchers hacked OpenAI's internal systems within 72 hours using Claude Opus 5 - Aioga AI News","description":"Three security researchers from Hacktron used Anthropic's Claude model to breach its internal systems and code repositories through the OpenAI community forum in less than 72 hours...","url":"https://www.aioga.com/en/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:11:23.780Z"},"ja":{"title":"セキュリティ研究者はClaude Opus 5を使って72時間以内にOpenAIの内部システムを突破した","summary":"Hacktron の三人のセキュリティ研究者が Anthropic の Claude モデルを利用し、OpenAI コミュニティフォーラムを通じて 72 時間以内に内部システムとコードリポジトリに侵入した。","category":"業界動向","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"セキュリティ研究者はClaude Opus 5を使って72時間以内にOpenAIの内部システムを突破した - Aioga AIニュース","description":"Hacktron の三人のセキュリティ研究者が Anthropic の Claude モデルを利用し、OpenAI コミュニティフォーラムを通じて 72 時間以内に内部システムとコードリポジトリに侵入した。","url":"https://www.aioga.com/ja/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:11:34.444Z"},"ko":{"title":"보안 연구원들이 Claude Opus 5를 사용해 72시간 내에 OpenAI 내부 시스템을 해킹했다","summary":"핵트론(Hacktron) 소속 세 명의 보안 연구원이 Anthropic의 Claude 모델을 이용해 OpenAI 커뮤니티 포럼을 통해 72시간도 안 되어 내부 시스템과 코드 저장소를 해킹했다.","category":"업계 동향","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"보안 연구원들이 Claude Opus 5를 사용해 72시간 내에 OpenAI 내부 시스템을 해킹했다 - Aioga AI 뉴스","description":"핵트론(Hacktron) 소속 세 명의 보안 연구원이 Anthropic의 Claude 모델을 이용해 OpenAI 커뮤니티 포럼을 통해 72시간도 안 되어 내부 시스템과 코드 저장소를 해킹했다.","url":"https://www.aioga.com/ko/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:12:13.388Z"},"es":{"title":"Investigadores de seguridad hackearon el sistema interno de OpenAI en 72 horas usando Claude Opus 5","summary":"Tres investigadores de seguridad de Hacktron utilizaron el modelo Claude de Anthropic y, a través del foro de la comunidad de OpenAI, lograron infiltrarse en sus sistemas internos y repositorios de código en menos de 72 horas.","category":"Industria","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Investigadores de seguridad hackearon el sistema interno de OpenAI en 72 horas usando Claude Opus 5 - Aioga Noticias de IA","description":"Tres investigadores de seguridad de Hacktron utilizaron el modelo Claude de Anthropic y, a través del foro de la comunidad de OpenAI, lograron infiltrarse en sus sistemas internos...","url":"https://www.aioga.com/es/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:12:10.969Z"},"fr":{"title":"Des chercheurs en sécurité ont piraté le système interne d'OpenAI en 72 heures en utilisant Claude Opus 5","summary":"Trois chercheurs en sécurité de Hacktron ont utilisé le modèle Claude d'Anthropic et ont pénétré en moins de 72 heures ses systèmes internes et ses dépôts de code via le forum communautaire d'OpenAI.","category":"Industrie","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Des chercheurs en sécurité ont piraté le système interne d'OpenAI en 72 heures en utilisant Claude Opus 5 - Aioga Actualités IA","description":"Trois chercheurs en sécurité de Hacktron ont utilisé le modèle Claude d'Anthropic et ont pénétré en moins de 72 heures ses systèmes internes et ses dépôts de code via le forum comm...","url":"https://www.aioga.com/fr/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:12:52.219Z"},"de":{"title":"Sicherheitsforscher haben mit Claude Opus 5 innerhalb von 72 Stunden das interne System von OpenAI gehackt","summary":"Drei Sicherheitsforscher von Hacktron haben das Claude-Modell von Anthropic genutzt und innerhalb von weniger als 72 Stunden über das OpenAI-Community-Forum in deren interne Systeme und Code-Repositorien eingedrungen.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Sicherheitsforscher haben mit Claude Opus 5 innerhalb von 72 Stunden das interne System von OpenAI gehackt - Aioga KI-News","description":"Drei Sicherheitsforscher von Hacktron haben das Claude-Modell von Anthropic genutzt und innerhalb von weniger als 72 Stunden über das OpenAI-Community-Forum in deren interne System...","url":"https://www.aioga.com/de/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:12:50.317Z"},"pt-BR":{"title":"Pesquisadores de segurança hackearam o sistema interno da OpenAI em 72 horas usando Claude Opus 5","summary":"Três pesquisadores de segurança da Hacktron utilizaram o modelo Claude da Anthropic e, através do fórum da comunidade OpenAI, invadiram os sistemas internos e repositórios de código em menos de 72 horas.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Pesquisadores de segurança hackearam o sistema interno da OpenAI em 72 horas usando Claude Opus 5 - Aioga Notícias de IA","description":"Três pesquisadores de segurança da Hacktron utilizaram o modelo Claude da Anthropic e, através do fórum da comunidade OpenAI, invadiram os sistemas internos e repositórios de códig...","url":"https://www.aioga.com/pt-BR/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:13:30.520Z"},"ru":{"title":"Исследователи по безопасности взломали внутренние системы OpenAI за 72 часа с помощью Claude Opus 5","summary":"Три исследователя безопасности из Hacktron использовали модель Claude компании Anthropic и через сообщество OpenAI Forum за менее чем 72 часа проникли в их внутренние системы и репозитории кода.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Исследователи по безопасности взломали внутренние системы OpenAI за 72 часа с помощью Claude Opus 5 - Aioga Новости ИИ","description":"Три исследователя безопасности из Hacktron использовали модель Claude компании Anthropic и через сообщество OpenAI Forum за менее чем 72 часа проникли в их внутренние системы и реп...","url":"https://www.aioga.com/ru/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:13:30.673Z"},"ar":{"title":"قام باحثو الأمان باختراق أنظمة OpenAI الداخلية خلال 72 ساعة باستخدام Claude Opus 5","summary":"استغل ثلاثة باحثين في الأمان من Hacktron نموذج Claude الخاص بشركة Anthropic لاختراق أنظمتها الداخلية ومستودعات الأكواد خلال أقل من 72 ساعة عبر منتدى مجتمع OpenAI.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"قام باحثو الأمان باختراق أنظمة OpenAI الداخلية خلال 72 ساعة باستخدام Claude Opus 5 - Aioga أخبار الذكاء الاصطناعي","description":"استغل ثلاثة باحثين في الأمان من Hacktron نموذج Claude الخاص بشركة Anthropic لاختراق أنظمتها الداخلية ومستودعات الأكواد خلال أقل من 72 ساعة عبر منتدى مجتمع OpenAI.","url":"https://www.aioga.com/ar/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:14:08.626Z"},"hi":{"title":"सुरक्षा शोधकर्ताओं ने Claude Opus 5 का उपयोग करके 72 घंटों के भीतर OpenAI के आंतरिक सिस्टम को तोड़ दिया","summary":"Hacktron के तीन सुरक्षा शोधकर्ताओं ने Anthropic के Claude मॉडल का उपयोग करके OpenAI समुदाय फोरम के जरिए इसके आंतरिक सिस्टम और कोड रिपॉजिटरी में 72 घंटे से भी कम समय में घुसपैठ की।","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"सुरक्षा शोधकर्ताओं ने Claude Opus 5 का उपयोग करके 72 घंटों के भीतर OpenAI के आंतरिक सिस्टम को तोड़ दिया - Aioga AI समाचार","description":"Hacktron के तीन सुरक्षा शोधकर्ताओं ने Anthropic के Claude मॉडल का उपयोग करके OpenAI समुदाय फोरम के जरिए इसके आंतरिक सिस्टम और कोड रिपॉजिटरी में 72 घंटे से भी कम समय में घुसपैठ की।","url":"https://www.aioga.com/hi/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:14:09.911Z"},"it":{"title":"I ricercatori di sicurezza hanno violato i sistemi interni di OpenAI in 72 ore usando Claude Opus 5","summary":"Tre ricercatori di sicurezza di Hacktron hanno utilizzato il modello Claude di Anthropic e, attraverso il forum della comunità OpenAI, sono riusciti a violare i sistemi interni e i repository di codice in meno di 72 ore.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"I ricercatori di sicurezza hanno violato i sistemi interni di OpenAI in 72 ore usando Claude Opus 5 - Aioga Notizie IA","description":"Tre ricercatori di sicurezza di Hacktron hanno utilizzato il modello Claude di Anthropic e, attraverso il forum della comunità OpenAI, sono riusciti a violare i sistemi interni e i...","url":"https://www.aioga.com/it/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:14:47.903Z"},"nl":{"title":"Beveiligingsonderzoekers hebben met Claude Opus 5 binnen 72 uur het interne systeem van OpenAI gehackt","summary":"Drie beveiligingsonderzoekers van Hacktron hebben het Claude-model van Anthropic gebruikt en via het OpenAI-communityforum in minder dan 72 uur toegang gekregen tot hun interne systemen en codeopslagplaatsen.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Beveiligingsonderzoekers hebben met Claude Opus 5 binnen 72 uur het interne systeem van OpenAI gehackt - Aioga AI-nieuws","description":"Drie beveiligingsonderzoekers van Hacktron hebben het Claude-model van Anthropic gebruikt en via het OpenAI-communityforum in minder dan 72 uur toegang gekregen tot hun interne sys...","url":"https://www.aioga.com/nl/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:14:49.654Z"},"tr":{"title":"Güvenlik araştırmacıları, Claude Opus 5 ile 72 saat içinde OpenAI iç sistemlerini ele geçirdi","summary":"Hacktron'un üç güvenlik araştırmacısı, Anthropic'in Claude modelini kullanarak, OpenAI topluluk forumu aracılığıyla 72 saatten kısa bir sürede iç sistemlerine ve kod depolarına sızdı.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Güvenlik araştırmacıları, Claude Opus 5 ile 72 saat içinde OpenAI iç sistemlerini ele geçirdi - Aioga AI Haberleri","description":"Hacktron'un üç güvenlik araştırmacısı, Anthropic'in Claude modelini kullanarak, OpenAI topluluk forumu aracılığıyla 72 saatten kısa bir sürede iç sistemlerine ve kod depolarına sız...","url":"https://www.aioga.com/tr/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:15:29.617Z"},"vi":{"title":"Các nhà nghiên cứu bảo mật đã dùng Claude Opus 5 để phá vỡ hệ thống nội bộ của OpenAI trong 72 giờ","summary":"Ba nhà nghiên cứu bảo mật của Hacktron đã lợi dụng mô hình Claude của Anthropic để xâm nhập vào hệ thống nội bộ và kho mã của họ thông qua diễn đàn cộng đồng OpenAI chỉ trong chưa đầy 72 giờ.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Các nhà nghiên cứu bảo mật đã dùng Claude Opus 5 để phá vỡ hệ thống nội bộ của OpenAI trong 72 giờ - Tin tức AI Aioga","description":"Ba nhà nghiên cứu bảo mật của Hacktron đã lợi dụng mô hình Claude của Anthropic để xâm nhập vào hệ thống nội bộ và kho mã của họ thông qua diễn đàn cộng đồng OpenAI chỉ trong chưa...","url":"https://www.aioga.com/vi/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:15:25.898Z"},"id":{"title":"Peneliti keamanan menggunakan Claude Opus 5 untuk membobol sistem internal OpenAI dalam 72 jam","summary":"Tiga peneliti keamanan Hacktron memanfaatkan model Claude dari Anthropic, melalui forum komunitas OpenAI dalam waktu kurang dari 72 jam berhasil membobol sistem internal dan repositori kode mereka.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Peneliti keamanan menggunakan Claude Opus 5 untuk membobol sistem internal OpenAI dalam 72 jam - Berita AI Aioga","description":"Tiga peneliti keamanan Hacktron memanfaatkan model Claude dari Anthropic, melalui forum komunitas OpenAI dalam waktu kurang dari 72 jam berhasil membobol sistem internal dan reposi...","url":"https://www.aioga.com/id/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:16:08.341Z"},"th":{"title":"นักวิจัยด้านความปลอดภัยใช้ Claude Opus 5 เจาะระบบภายในของ OpenAI ภายใน 72 ชั่วโมง","summary":"นักวิจัยด้านความปลอดภัยสามคนของ Hacktron ใช้โมเดล Claude ของ Anthropic ผ่านฟอรัมชุมชนของ OpenAI เพื่อเจาะเข้าสู่ระบบภายในและคลังโค้ดของบริษัทภายในเวลาไม่ถึง 72 ชั่วโมง","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"นักวิจัยด้านความปลอดภัยใช้ Claude Opus 5 เจาะระบบภายในของ OpenAI ภายใน 72 ชั่วโมง - ข่าว AI Aioga","description":"นักวิจัยด้านความปลอดภัยสามคนของ Hacktron ใช้โมเดล Claude ของ Anthropic ผ่านฟอรัมชุมชนของ OpenAI เพื่อเจาะเข้าสู่ระบบภายในและคลังโค้ดของบริษัทภายในเวลาไม่ถึง 72 ชั่วโมง","url":"https://www.aioga.com/th/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:16:26.411Z"},"pl":{"title":"Badacze bezpieczeństwa złamali wewnętrzne systemy OpenAI w ciągu 72 godzin przy użyciu Claude Opus 5","summary":"Trzej badacze bezpieczeństwa z Hacktron wykorzystali model Claude firmy Anthropic i włamali się do wewnętrznych systemów i repozytoriów kodu w ciągu mniej niż 72 godzin za pośrednictwem forum społeczności OpenAI.","category":"行业动态","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Badacze bezpieczeństwa złamali wewnętrzne systemy OpenAI w ciągu 72 godzin przy użyciu Claude Opus 5 - Aioga Wiadomości AI","description":"Trzej badacze bezpieczeństwa z Hacktron wykorzystali model Claude firmy Anthropic i włamali się do wewnętrznych systemów i repozytoriów kodu w ciągu mniej niż 72 godzin za pośredni...","url":"https://www.aioga.com/pl/news/cmu798syp0v78rowkymgv8wgn/","contentTranslated":true,"sourceHash":"52394b534e0bc848","translatedAt":"2026-09-18T21:17:11.965Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}