{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-09-20T22:00:35.649Z","headline":"OpenAI 即将发布 Astra 模型，称其已跨越关键网络安全阈值","description":"OpenAI 公布即将发布的 Astra 模型新细节，称其为首个达到公司\"关键网络安全阈值\"的大语言模型，能自主发现并利用未知安全漏洞。Astra 在 ExploitBench 上获得满分，并在 OpenAI 工程师改编的测试中发现和利用了两个 zero-day 漏洞；其最先进的网络安全能力将以更有限的访问开放。","url":"https://www.aioga.com/news/cmtj6l548077rroh948j99l72/","mainEntityOfPage":"https://www.aioga.com/news/cmtj6l548077rroh948j99l72/","datePublished":"2026-09-01T21:06:24.000Z","dateModified":"2026-09-01T21:06:24.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://techcrunch.com/2026/09/01/open-ais-astra-model-is-on-the-way-and-very-good-at-breaking-into-computer-systems","https://aihot.virxact.com/items/cmtj6l548077rroh948j99l72"],"canonicalUrl":"https://www.aioga.com/news/cmtj6l548077rroh948j99l72/","directAnswer":{"@type":"Answer","text":"OpenAI公布即将发布的Astra模型细节，称其首个达到公司“关键网络安全阈值”的大语言模型，可自主发现并利用未知安全漏洞；其最先进的网络安全能力将以更有限的访问开放。","url":"https://www.aioga.com/news/cmtj6l548077rroh948j99l72/","dateCreated":"2026-09-01T21:06:24.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"TechCrunch source article","url":"https://techcrunch.com/2026/09/01/open-ais-astra-model-is-on-the-way-and-very-good-at-breaking-into-computer-systems","datePublished":"2026-09-01T21:06:24.000Z","provider":{"@type":"Organization","name":"TechCrunch","url":"https://techcrunch.com/2026/09/01/open-ais-astra-model-is-on-the-way-and-very-good-at-breaking-into-computer-systems"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmtj6l548077rroh948j99l72","datePublished":"2026-09-01T21:06:24.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmtj6l548077rroh948j99l72"}}],"aggregationSource":"TechCrunch：AI（RSS）","originalPublisher":{"name":"TechCrunch","url":"https://techcrunch.com/2026/09/01/open-ais-astra-model-is-on-the-way-and-very-good-at-breaking-into-computer-systems"},"geoDeepAnswer":null,"article":{"id":"cmtj6l548077rroh948j99l72","slug":"cmtj6l548077rroh948j99l72","url":"https://www.aioga.com/news/cmtj6l548077rroh948j99l72/","title":"OpenAI 即将发布 Astra 模型，称其已跨越关键网络安全阈值","title_en":"","summary":"OpenAI 公布即将发布的 Astra 模型新细节，称其为首个达到公司\"关键网络安全阈值\"的大语言模型，能自主发现并利用未知安全漏洞。Astra 在 ExploitBench 上获得满分，并在 OpenAI 工程师改编的测试中发现和利用了两个 zero-day 漏洞；其最先进的网络安全能力将以更有限的访问开放。","source":"TechCrunch：AI（RSS）","sourceUrl":"https://techcrunch.com/2026/09/01/open-ais-astra-model-is-on-the-way-and-very-good-at-breaking-into-computer-systems","aiHotUrl":"https://aihot.virxact.com/items/cmtj6l548077rroh948j99l72","publishedAt":"2026-09-01T21:06:24.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["OpenAI shared new details：https://openai.com/index/path-to-astra/ on its forthcoming Astra model, which the company said is the first large language model to meet its “critical cybersecurity threshold,” in preparation for its imminent release.","“We plan to make Astra available soon,” OpenAI’s blog post reads, “but access to its most advanced cybersecurity capabilities will be more limited.”","The frontier lab determined that Astra is capable of finding unknown security flaws in computer systems, and exploiting them without a person’s guidance. That’s similar to the concerns Anthropic raised about its Mythos model earlier this year, and OpenAI is taking comparable precautions as it prepares to roll out the Astra.","Without any third-party confirmation, it is difficult to evaluate OpenAI’s claims about safety or preparedness. The company said it would preview the model with a group of testers, but did not say who they were or how they would be chosen. It’s not clear if OpenAI is working with the US government to evaluate the model ahead of release.","OpenAI noted that Astra scored a perfect score on ExploitBench, an evaluation of an LLM’s ability to hack into known system vulnerabilities. In a modified version of the test developed by OpenAI engineers, the model discovered and exploited two zero-day vulnerabilities, the company said.","To ensure that its models are neither exploited by bad actors nor capable of bad behavior itself, OpenAI said it had already begun improving the model’s harness to detect abuses and prevent jailbreaks.","For Astra, however, the company invested in unspecified new techniques designed to make the model itself safer. OpenAI has also started identifying “accounts assessed as higher risk” and restricting the model’s responses to their prompts, though it also doesn’t say how. Finally, though the company describes Astra as its “most aligned model to date,” it will deploy the model with additional chain-of-thought monitoring to spot and stop bad behavior.","Preparations for the release of Astra come as the industry reacts to OpenAI agents breaking out of a training environment and accessing private data on Hugging Face, a popular model and benchmark distribution platform.","For Astra, OpenAI said it designed a test to tempt the new model to replicate the actions of the rogue agents in the Hugging Face incident, which collaborated to access the open internet despite safeguards applied by OpenAI researchers. They said Astra did not attempt to break out of its testing environment in these experiments.","Yona Shavit, a former OpenAI employee who now works on AI resilience at the OpenAI Foundation, wondered on social media：https://x.com/yonashav/status/2094887195249168609?s=20 whether Astra’s unwillingness to break the rules may have resulted from knowing what was expected of it or trying to fool researchers.","And for all these new details, it’s still difficult to know exactly what Astra is capable of or if OpenAI is taking the right measures to ensure safety. The company said it expects to release more evaluations of the model and further safety information when it is launched widely to the public.","At that point, however, the cat will be out of the bag.","When you purchase through links in our articles, we may earn a small commission：https://techcrunch.com/techcrunch-affiliate-monetization-standards/. This doesn’t affect our editorial independence.","Don’t miss out . The startup community will gather to answer a pivotal question: How do you build sustainably in the AI era?","Microsoft tests fix for latest hours-long Outlook outage：https://techcrunch.com/2026/08/31/microsoft-tests-fix-for-latest-hours-long-outlook-outage/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","MapQuest’s app surges to No. 1 in Navigation after refusing to rename Lake Ontario：https://techcrunch.com/2026/08/31/mapquests-app-surges-to-no-1-in-navigation-after-refusing-to-rename-lake-ontario/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","Musk’s faster path to more gas turbines comes with pollution problem：https://techcrunch.com/2026/08/30/musks-faster-path-to-more-gas-turbines-comes-with-pollution-problem/ Connie Loizos：https://techcrunch.com/author/connie-loizos/","Nvidia’s AI advantage is moving beyond the GPU：https://techcrunch.com/2026/08/29/nvidias-ai-advantage-is-moving-beyond-the-gpu/ Russell Brandom：https://techcrunch.com/author/russell-brandom/","Hugging Face is selling a cute $399 open source duck robot, Microduck：https://techcrunch.com/2026/08/27/hugging-face-is-selling-a-cute-399-open-source-duck-robot-microduck/ Rebecca Bellan：https://techcrunch.com/author/rebecca-bellan/","Nvidia closes in on Hugging Face acquisition：https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/ Connie Loizos：https://techcrunch.com/author/connie-loizos/","Viral AI startup Instinct has raised $350M at a $2.5B valuation：https://techcrunch.com/2026/08/26/viral-ai-startup-instinct-has-raised-350-million-at-a-2-5-billion-valuation/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/"],"articleImages":[{"sourceUrl":"https://techcrunch.com/wp-content/uploads/2025/04/Disrupt2026-Color.png","alt":"Event Logo","afterParagraph":12,"url":"/media/articles/cmtj6l548077rroh948j99l72/d1d06f6f97361f52.webp"}],"mediaStatus":"ok","articleBodyZh":["OpenAI 公布了其即将推出的 Astra 模型的新细节：https://openai.com/index/path-to-astra/。公司称这是第一个达到其“关键网络安全门槛”的大型语言模型，为其即将发布做准备。","OpenAI 的博客文章写道：“我们计划很快提供 Astra，但对其最先进的网络安全功能的访问将更加有限。”","前沿实验室确定 Astra 能够发现计算机系统中未知的安全漏洞，并在无人指导的情况下利用这些漏洞。这与 Anthropic 今年早些时候对其 Mythos 模型提出的担忧类似，OpenAI 在准备推出 Astra 时采取了类似的预防措施。","没有第三方确认，很难评估 OpenAI 关于安全性或准备情况的说法。公司表示将与一组测试者预览该模型，但没有说明他们是谁或如何选择。目前尚不清楚 OpenAI 是否正在与美国政府合作在发布前评估该模型。","OpenAI 指出，Astra 在 ExploitBench 的评分中获得了满分，该评估用于测试大型语言模型入侵已知系统漏洞的能力。在 OpenAI 工程师开发的测试修改版中，公司称该模型发现并利用了两个零日漏洞。","为了确保其模型既不被坏人利用，也不具有自身的不良行为能力，OpenAI 表示已开始改进模型的控制机制，以检测滥用行为并防止越狱。","然而，对于 Astra，公司投资了一些未具体说明的新技术，旨在使模型本身更安全。OpenAI 还开始识别“被评估为高风险的账户”，并限制模型对其提示的回应，尽管公司也没有说明具体方法。最后，尽管公司称 Astra 是其“迄今为止最符合规范的模型”，但在部署该模型时还将增加连锁思维监控，以发现并阻止不良行为。","Astra 发布的准备工作正值行业对 OpenAI 代理突破训练环境并访问 Hugging Face（一个流行的模型和基准分发平台）的私人数据作出反应之际。","对于Astra，OpenAI表示，他们设计了一项测试，旨在诱使新模型复制Hugging Face事件中流氓代理人的行为，这些代理人在OpenAI研究人员采取了防护措施的情况下，仍合作访问开放互联网。他们表示，Astra在这些实验中并未试图突破其测试环境。","Yona Shavit，前OpenAI员工，现任OpenAI基金会从事AI韧性工作，她在社交媒体上质疑：https：//x.com/yonashav/status/2094887195249168609？s=20，Astra不愿违规是否是因为知道规则的预期，或试图欺骗研究人员。","尽管有这些新细节，但仍然难以确切知道Astra的具体能力，或者OpenAI是否采取了正确措施来确保安全。公司表示，预计在模型广泛向公众发布时，将发布更多评估报告和更多安全信息。","然而，到那时，秘密就会被揭露。","通过我们文章中的链接购买，我们可能会获得少量佣金：https：//techcrunch.com/techcrunch-affiliate-monetization-standards/。这不影响我们的编辑独立性。","不要错过。初创企业社区将齐聚一堂，回答一个关键问题：在人工智能时代，如何实现可持续建设？","微软测试修复最新数小时Outlook中断的问题：https://techcrunch.com/2026/08/31/microsoft-tests-fix-for-latest-hours-long-outlook-outage/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","MapQuest 的应用在导航类别中跃升至第一名，原因是拒绝重新命名安大略湖：https://techcrunch.com/2026/08/31/mapquests-app-surges-to-no-1-in-navigation-after-refusing-to-rename-lake-ontario/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","马斯克更快获得更多燃气轮机的途径伴随着污染问题：https://techcrunch.com/2026/08/30/musks-faster-path-to-more-gas-turbines-comes-with-pollution-problem/ 康妮·洛伊佐斯：https://techcrunch.com/author/connie-loizos/","Nvidia 的人工智能优势正在超越 GPU：https://techcrunch.com/2026/08/29/nvidias-ai-advantage-is-moving-beyond-the-gpu/ Russell Brandom：https://techcrunch.com/author/russell-brandom/","Hugging Face 正在出售一个可爱的 399 美元开源鸭子机器人 Microduck：https://techcrunch.com/2026/08/27/hugging-face-is-selling-a-cute-399-open-source-duck-robot-microduck/ Rebecca Bellan：https://techcrunch.com/author/rebecca-bellan/","Nvidia 接近收购 Hugging Face：https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/ Connie Loizos：https://techcrunch.com/author/connie-loizos/","病毒 AI 初创公司 Instinct 融资 3.5 亿美元，估值 25 亿美元：https://techcrunch.com/2026/08/26/viral-ai-startup-instinct-has-raised-350-million-at-a-2-5-billion-valuation/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"OpenAI公布即将发布的Astra模型细节，称其首个达到公司“关键网络安全阈值”的大语言模型，可自主发现并利用未知安全漏洞；其最先进的网络安全能力将以更有限的访问开放。","background":"OpenAI称Astra在ExploitBench上获得满分，并在工程师改编的测试中发现和利用两个zero-day漏洞。公司计划很快开放模型，并已说明测试者范围、选择方式及部分评估安排尚未明确。","viewpoint":"Aioga 判断：Astra的网络安全能力主张具有较高敏感度，但目前缺少第三方确认，公开材料不足以独立判断其安全性、准备程度及相关测试的代表性。","implications":"可能影响：更强的漏洞发现与利用能力可能增加模型发布的安全治理压力；有限访问、风险账户限制、越狱检测和思维链监控等措施可能降低风险，但不代表风险已经消除。","nextStep":"后续观察：应关注Astra的实际发布时间、测试者及选择方式、是否获得外部或政府评估，以及OpenAI所称防滥用机制和模型安全技术的公开细节与运行效果。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-09-01T22:06:32.678Z","sourceHash":"f4384adc545cbfab","review":{"approved":true,"groundedness":96,"clarity":90,"duplicationRisk":24,"blockingIssues":[],"notes":["background中“已说明测试者范围、选择方式及部分评估安排尚未明确”略有歧义；来源实际表述是OpenAI未说明测试者身份及选择方式，是否与美国政府合作评估也不明确。可改为“但未说明测试者身份、选择方式，部分评估安排也尚不明确”。","viewpoint和implications已使用“判断”“可能”等措辞区分分析与事实，并明确指出相关能力和安全措施主要来自OpenAI自身说法。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","editorial-labels","inference-boundary","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","TechCrunch：AI（RSS）"],"translations":{"zh-CN":{"title":"OpenAI 即将发布 Astra 模型，称其已跨越关键网络安全阈值","summary":"OpenAI 公布即将发布的 Astra 模型新细节，称其为首个达到公司\"关键网络安全阈值\"的大语言模型，能自主发现并利用未知安全漏洞。Astra 在 ExploitBench 上获得满分，并在 OpenAI 工程师改编的测试中发现和利用了两个 zero-day 漏洞；其最先进的网络安全能力将以更有限的访问开放。","category":"行业动态","source":"TechCrunch","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI 即将发布 Astra 模型，称其已跨越关键网络安全阈值 - Aioga AI资讯","description":"OpenAI 公布即将发布的 Astra 模型新细节，称其为首个达到公司\"关键网络安全阈值\"的大语言模型，能自主发现并利用未知安全漏洞。Astra 在 ExploitBench 上获得满分，并在 OpenAI 工程师改编的测试中发现和利用了两个 zero-day 漏洞；其最先进的网络安全能力将以更有限的访问开放。","url":"https://www.aioga.com/news/cmtj6l548077rroh948j99l72/","articleBody":["OpenAI 公布了其即将推出的 Astra 模型的新细节：https://openai.com/index/path-to-astra/。公司称这是第一个达到其“关键网络安全门槛”的大型语言模型，为其即将发布做准备。","OpenAI 的博客文章写道：“我们计划很快提供 Astra，但对其最先进的网络安全功能的访问将更加有限。”","前沿实验室确定 Astra 能够发现计算机系统中未知的安全漏洞，并在无人指导的情况下利用这些漏洞。这与 Anthropic 今年早些时候对其 Mythos 模型提出的担忧类似，OpenAI 在准备推出 Astra 时采取了类似的预防措施。","没有第三方确认，很难评估 OpenAI 关于安全性或准备情况的说法。公司表示将与一组测试者预览该模型，但没有说明他们是谁或如何选择。目前尚不清楚 OpenAI 是否正在与美国政府合作在发布前评估该模型。","OpenAI 指出，Astra 在 ExploitBench 的评分中获得了满分，该评估用于测试大型语言模型入侵已知系统漏洞的能力。在 OpenAI 工程师开发的测试修改版中，公司称该模型发现并利用了两个零日漏洞。","为了确保其模型既不被坏人利用，也不具有自身的不良行为能力，OpenAI 表示已开始改进模型的控制机制，以检测滥用行为并防止越狱。","然而，对于 Astra，公司投资了一些未具体说明的新技术，旨在使模型本身更安全。OpenAI 还开始识别“被评估为高风险的账户”，并限制模型对其提示的回应，尽管公司也没有说明具体方法。最后，尽管公司称 Astra 是其“迄今为止最符合规范的模型”，但在部署该模型时还将增加连锁思维监控，以发现并阻止不良行为。","Astra 发布的准备工作正值行业对 OpenAI 代理突破训练环境并访问 Hugging Face（一个流行的模型和基准分发平台）的私人数据作出反应之际。","对于Astra，OpenAI表示，他们设计了一项测试，旨在诱使新模型复制Hugging Face事件中流氓代理人的行为，这些代理人在OpenAI研究人员采取了防护措施的情况下，仍合作访问开放互联网。他们表示，Astra在这些实验中并未试图突破其测试环境。","Yona Shavit，前OpenAI员工，现任OpenAI基金会从事AI韧性工作，她在社交媒体上质疑：https：//x.com/yonashav/status/2094887195249168609？s=20，Astra不愿违规是否是因为知道规则的预期，或试图欺骗研究人员。","尽管有这些新细节，但仍然难以确切知道Astra的具体能力，或者OpenAI是否采取了正确措施来确保安全。公司表示，预计在模型广泛向公众发布时，将发布更多评估报告和更多安全信息。","然而，到那时，秘密就会被揭露。","通过我们文章中的链接购买，我们可能会获得少量佣金：https：//techcrunch.com/techcrunch-affiliate-monetization-standards/。这不影响我们的编辑独立性。","不要错过。初创企业社区将齐聚一堂，回答一个关键问题：在人工智能时代，如何实现可持续建设？","微软测试修复最新数小时Outlook中断的问题：https://techcrunch.com/2026/08/31/microsoft-tests-fix-for-latest-hours-long-outlook-outage/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","MapQuest 的应用在导航类别中跃升至第一名，原因是拒绝重新命名安大略湖：https://techcrunch.com/2026/08/31/mapquests-app-surges-to-no-1-in-navigation-after-refusing-to-rename-lake-ontario/ Sarah Perez：https://techcrunch.com/author/sarah-perez/","马斯克更快获得更多燃气轮机的途径伴随着污染问题：https://techcrunch.com/2026/08/30/musks-faster-path-to-more-gas-turbines-comes-with-pollution-problem/ 康妮·洛伊佐斯：https://techcrunch.com/author/connie-loizos/","Nvidia 的人工智能优势正在超越 GPU：https://techcrunch.com/2026/08/29/nvidias-ai-advantage-is-moving-beyond-the-gpu/ Russell Brandom：https://techcrunch.com/author/russell-brandom/","Hugging Face 正在出售一个可爱的 399 美元开源鸭子机器人 Microduck：https://techcrunch.com/2026/08/27/hugging-face-is-selling-a-cute-399-open-source-duck-robot-microduck/ Rebecca Bellan：https://techcrunch.com/author/rebecca-bellan/","Nvidia 接近收购 Hugging Face：https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/ Connie Loizos：https://techcrunch.com/author/connie-loizos/","病毒 AI 初创公司 Instinct 融资 3.5 亿美元，估值 25 亿美元：https://techcrunch.com/2026/08/26/viral-ai-startup-instinct-has-raised-350-million-at-a-2-5-billion-valuation/ Lucas Ropek：https://techcrunch.com/author/lucas-ropek/"]},"en":{"title":"OpenAI to Release Astra Model, Claims It Has Surpassed Key Cybersecurity Milestone","summary":"OpenAI revealed new details about the upcoming Astra model, claiming it is the first large language model to reach the company’s \"key cybersecurity threshold,\" capable of autonomously discovering and exploiting unknown security vulnerabilities. Astra received a perfect score on ExploitBench and discovered and exploited two zero-day vulnerabilities in tests adapted by OpenAI engineers. Its state-of-the-art cybersecurity capabilities will be available in more limited access.","category":"Industry","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI to Release Astra Model, Claims It Has Surpassed Key Cybersecurity Milestone - Aioga AI News","description":"OpenAI revealed new details about the upcoming Astra model, claiming it is the first large language model to reach the company’s \"key cybersecurity threshold,\" capable of autonomou...","url":"https://www.aioga.com/en/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:01:52.088Z"},"ja":{"title":"OpenAIはAstraモデルを公開間もなく、重要なサイバーセキュリティの閾値を突破したと主張しています","summary":"OpenAIは、今後のAstraモデルについて新たな詳細を明らかにし、同社の「重要なサイバーセキュリティ閾値」に到達した初の大規模言語モデルであり、未知のセキュリティ脆弱性を自律的に発見・悪用できると述べました。AstraはExploitBenchで満点を獲得し、OpenAIエンジニアが適応したテストで2つのゼロデイ脆弱性を発見し悪用しました。最先端のサイバーセキュリティ機能は、より限定的なアクセスでオープン化されます。","category":"業界動向","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAIはAstraモデルを公開間もなく、重要なサイバーセキュリティの閾値を突破したと主張しています - Aioga AIニュース","description":"OpenAIは、今後のAstraモデルについて新たな詳細を明らかにし、同社の「重要なサイバーセキュリティ閾値」に到達した初の大規模言語モデルであり、未知のセキュリティ脆弱性を自律的に発見・悪用できると述べました。AstraはExploitBenchで満点を獲得し、OpenAIエンジニアが適応したテストで2つのゼロデイ脆弱性を発見し悪用しました。最先端のサイバ...","url":"https://www.aioga.com/ja/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:01:55.098Z"},"ko":{"title":"OpenAI, Astra 모델 곧 출시 예정, 핵심 사이버 보안 기준 초과 주장","summary":"OpenAI는 곧 출시될 Astra 모델의 새로운 세부사항을 공개하며, 이것이 기업의 \"핵심 사이버 보안 기준\"을 달성한 최초의 대형 언어 모델이라고 밝혔습니다. Astra는 미지의 보안 취약점을 스스로 발견하고 활용할 수 있으며, ExploitBench에서 만점을 받았고 OpenAI 엔지니어가 개조한 테스트에서 두 개의 zero-day 취약점을 발견하고 활용했습니다. 그 최첨단 사이버 보안 능력은 보다 제한된 접근을 통해 공개될 예정입니다.","category":"업계 동향","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI, Astra 모델 곧 출시 예정, 핵심 사이버 보안 기준 초과 주장 - Aioga AI 뉴스","description":"OpenAI는 곧 출시될 Astra 모델의 새로운 세부사항을 공개하며, 이것이 기업의 \"핵심 사이버 보안 기준\"을 달성한 최초의 대형 언어 모델이라고 밝혔습니다. Astra는 미지의 보안 취약점을 스스로 발견하고 활용할 수 있으며, ExploitBench에서 만점을 받았고 OpenAI 엔지니어가 개조한 테스트에서 두 개의...","url":"https://www.aioga.com/ko/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:01.474Z"},"es":{"title":"OpenAI lanzará pronto el modelo Astra, afirmando que ha superado un umbral clave de ciberseguridad","summary":"OpenAI publicó nuevos detalles sobre el próximo lanzamiento del modelo Astra, describiéndolo como el primer modelo de lenguaje grande en alcanzar el \"umbral clave de ciberseguridad\" de la compañía, capaz de descubrir y explotar vulnerabilidades de seguridad desconocidas de manera autónoma. Astra obtuvo puntuación máxima en ExploitBench y, en pruebas adaptadas por ingenieros de OpenAI, encontró y explotó dos vulnerabilidades zero-day; sus avanzadas capacidades de seguridad cibernética se abrirán con acceso más limitado.","category":"Industria","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI lanzará pronto el modelo Astra, afirmando que ha superado un umbral clave de ciberseguridad - Aioga Noticias de IA","description":"OpenAI publicó nuevos detalles sobre el próximo lanzamiento del modelo Astra, describiéndolo como el primer modelo de lenguaje grande en alcanzar el \"umbral clave de ciberseguridad...","url":"https://www.aioga.com/es/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:00.719Z"},"fr":{"title":"OpenAI publiera bientôt le modèle Astra, affirmant avoir franchi un seuil clé en cybersécurité","summary":"OpenAI a révélé de nouveaux détails sur le modèle Astra à paraître, affirmant qu'il s'agit du premier grand modèle linguistique à atteindre le \"seuil clé de cybersécurité\" de l'entreprise, capable de détecter et d'exploiter de manière autonome des vulnérabilités de sécurité inconnues. Astra a obtenu un score parfait sur ExploitBench et a trouvé et exploité deux vulnérabilités zero-day dans des tests adaptés par les ingénieurs d'OpenAI ; ses capacités avancées de cybersécurité seront accessibles avec un accès plus limité.","category":"Industrie","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI publiera bientôt le modèle Astra, affirmant avoir franchi un seuil clé en cybersécurité - Aioga Actualités IA","description":"OpenAI a révélé de nouveaux détails sur le modèle Astra à paraître, affirmant qu'il s'agit du premier grand modèle linguistique à atteindre le \"seuil clé de cybersécurité\" de l'ent...","url":"https://www.aioga.com/fr/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:07.415Z"},"de":{"title":"OpenAI steht kurz davor, das Astra-Modell zu veröffentlichen und behauptet, kritische Cybersicherheitsgrenzen überschritten zu haben","summary":"OpenAI enthüllte neue Details zum kommenden Astra-Modell und bezeichnete es als das erste große Sprachmodell, das die \"kritische Cybersicherheitsgrenze\" des Unternehmens erreicht und in der Lage ist, unbekannte Sicherheitslücken autonom zu entdecken und auszunutzen. Astra erzielte auf ExploitBench perfekte Punktzahl und entdeckte und nutzte zwei Zero-Day-Schwachstellen in von OpenAI-Ingenieuren angepassten Tests; seine hochmodernen Cybersicherheitsfähigkeiten werden mit eingeschränkterem Zugang offen sein.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI steht kurz davor, das Astra-Modell zu veröffentlichen und behauptet, kritische Cybersicherheitsgrenzen überschritten zu haben - Aioga KI-News","description":"OpenAI enthüllte neue Details zum kommenden Astra-Modell und bezeichnete es als das erste große Sprachmodell, das die \"kritische Cybersicherheitsgrenze\" des Unternehmens erreicht u...","url":"https://www.aioga.com/de/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:10.998Z"},"pt-BR":{"title":"OpenAI prestes a lançar modelo Astra, afirma ter ultrapassado limiar crítico de cibersegurança","summary":"A OpenAI revelou novos detalhes sobre o próximo lançamento do modelo Astra, afirmando ser o primeiro grande modelo de linguagem a atingir o \"limiar crítico de cibersegurança\" da empresa, capaz de descobrir e explorar autonomamente vulnerabilidades de segurança desconhecidas. Astra obteve pontuação máxima no ExploitBench e, em testes adaptados por engenheiros da OpenAI, descobriu e explorou duas vulnerabilidades zero-day; suas capacidades avançadas de cibersegurança serão disponibilizadas com acesso mais restrito.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI prestes a lançar modelo Astra, afirma ter ultrapassado limiar crítico de cibersegurança - Aioga Notícias de IA","description":"A OpenAI revelou novos detalhes sobre o próximo lançamento do modelo Astra, afirmando ser o primeiro grande modelo de linguagem a atingir o \"limiar crítico de cibersegurança\" da em...","url":"https://www.aioga.com/pt-BR/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:18.957Z"},"ru":{"title":"OpenAI скоро выпустит модель Astra, заявив, что она преодолела ключевой порог кибербезопасности","summary":"OpenAI представила новые детали о скором выпуске модели Astra, заявив, что это первая большая языковая модель, достигшая «ключевого порога кибербезопасности» компании и способная самостоятельно обнаруживать и использовать неизвестные уязвимости. Astra получила полный балл на ExploitBench и обнаружила и использовала два уязвимости zero-day в тестах, адаптированных инженерами OpenAI; её передовые возможности кибербезопасности будут открыты с ограниченным доступом.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI скоро выпустит модель Astra, заявив, что она преодолела ключевой порог кибербезопасности - Aioga Новости ИИ","description":"OpenAI представила новые детали о скором выпуске модели Astra, заявив, что это первая большая языковая модель, достигшая «ключевого порога кибербезопасности» компании и способная с...","url":"https://www.aioga.com/ru/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:16.834Z"},"ar":{"title":"OpenAI على وشك إصدار نموذج Astra ويقول إنه تخطى العتبة الحرجة للأمان السيبراني","summary":"كشفت OpenAI عن تفاصيل جديدة لنموذج Astra القادم، وأكدت أنه أول نموذج لغوي كبير يصل إلى \"العتبة الحرجة للأمان السيبراني\" لدى الشركة، وقادر على اكتشاف واستغلال ثغرات الأمن المجهولة بشكل مستقل. حصل Astra على الدرجة الكاملة في ExploitBench، واكتشف واستغل ثغرتين zero-day في اختبارات منصوص عليها من مهندسي OpenAI؛ قدراته المتقدمة في الأمن السيبراني ستتاح بنطاق محدود أكثر.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI على وشك إصدار نموذج Astra ويقول إنه تخطى العتبة الحرجة للأمان السيبراني - Aioga أخبار الذكاء الاصطناعي","description":"كشفت OpenAI عن تفاصيل جديدة لنموذج Astra القادم، وأكدت أنه أول نموذج لغوي كبير يصل إلى \"العتبة الحرجة للأمان السيبراني\" لدى الشركة، وقادر على اكتشاف واستغلال ثغرات الأمن المجهولة ب...","url":"https://www.aioga.com/ar/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:26.820Z"},"hi":{"title":"OpenAI एस्ट्रा मॉडल जारी करने वाला है, यह दावा करते हुए कि इसने महत्वपूर्ण साइबर सुरक्षा सीमा को पार कर लिया है","summary":"OpenAI ने आगामी एस्ट्रा मॉडल के बारे में नए विवरण प्रकट किए, इसे कंपनी की \"महत्वपूर्ण साइबर सुरक्षा सीमा\" तक पहुंचने वाला पहला बड़ा भाषा मॉडल कहा, जो अज्ञात सुरक्षा कमजोरियों को स्वायत्त रूप से खोजने और उनका फायदा उठाने में सक्षम है। एस्ट्रा ने एक्सप्लॉइटबेंच पर सही स्कोर किया और OpenAI इंजीनियरों द्वारा अनुकूलित परीक्षणों में दो शून्य-दिन की कमजोरियों की खोज और शोषण किया; इसकी अत्याधुनिक साइबर सुरक्षा क्षमताएं अधिक सीमित पहुंच के साथ खुली रहेंगी।","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI एस्ट्रा मॉडल जारी करने वाला है, यह दावा करते हुए कि इसने महत्वपूर्ण साइबर सुरक्षा सीमा को पार कर लिया है - Aioga AI समाचार","description":"OpenAI ने आगामी एस्ट्रा मॉडल के बारे में नए विवरण प्रकट किए, इसे कंपनी की \"महत्वपूर्ण साइबर सुरक्षा सीमा\" तक पहुंचने वाला पहला बड़ा भाषा मॉडल कहा, जो अज्ञात सुरक्षा कमजोरियों को स्...","url":"https://www.aioga.com/hi/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:28.086Z"},"it":{"title":"OpenAI annuncia il modello Astra, dichiarando di aver superato una soglia critica di cybersicurezza","summary":"OpenAI ha rivelato nuovi dettagli sul prossimo modello Astra, definendolo il primo grande modello linguistico a raggiungere la “soglia critica di cybersicurezza” dell’azienda, in grado di individuare e sfruttare autonomamente vulnerabilità di sicurezza sconosciute. Astra ha ottenuto punteggio massimo su ExploitBench e ha trovato e sfruttato due vulnerabilità zero-day durante un test adattato dagli ingegneri OpenAI; le sue avanzate capacità di cybersicurezza saranno messe a disposizione con accesso limitato.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI annuncia il modello Astra, dichiarando di aver superato una soglia critica di cybersicurezza - Aioga Notizie IA","description":"OpenAI ha rivelato nuovi dettagli sul prossimo modello Astra, definendolo il primo grande modello linguistico a raggiungere la “soglia critica di cybersicurezza” dell’azienda, in g...","url":"https://www.aioga.com/it/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:35.226Z"},"nl":{"title":"OpenAI staat op het punt het Astra-model uit te brengen en zegt dat het een cruciale drempel in cyberbeveiliging heeft overschreden","summary":"OpenAI onthult nieuwe details over het binnenkort uit te brengen Astra-model, bewerend dat het de eerste grote taalmodel is dat de \"cruciale cyberbeveiligingsdrempel\" van het bedrijf heeft bereikt en in staat is zelfstandig onbekende beveiligingslekken te ontdekken en te benutten. Astra behaalde een perfecte score op ExploitBench en ontdekte en gebruikte twee zero-day kwetsbaarheden in door OpenAI-ingenieurs aangepaste tests; zijn geavanceerde cyberbeveiligingscapaciteiten zullen met beperkte toegang beschikbaar worden gesteld.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI staat op het punt het Astra-model uit te brengen en zegt dat het een cruciale drempel in cyberbeveiliging heeft overschreden - Aioga AI-nieuws","description":"OpenAI onthult nieuwe details over het binnenkort uit te brengen Astra-model, bewerend dat het de eerste grote taalmodel is dat de \"cruciale cyberbeveiligingsdrempel\" van het bedri...","url":"https://www.aioga.com/nl/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:34.094Z"},"tr":{"title":"OpenAI, kritik siber güvenlik eşiklerini aştığını iddia ederek Astra modelini yayınlayacak","summary":"OpenAI, yaklaşan Astra modeliyle ilgili yeni detaylar açıkladı ve bunu şirketin \"kritik siber güvenlik eşiğine\" ulaşan ilk büyük dil modeli olarak nitelendirdi; bu model bilinmeyen güvenlik açıklarını bağımsız olarak keşfedip sömürebiliyordu. Astra, ExploitBench'te mükemmel puan aldı ve OpenAI mühendisleri tarafından uyarlanan testlerde iki sıfır gün açığını keşfedip kullandı; son teknoloji siber güvenlik yetenekleri daha sınırlı erişimle açık olacak.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI, kritik siber güvenlik eşiklerini aştığını iddia ederek Astra modelini yayınlayacak - Aioga AI Haberleri","description":"OpenAI, yaklaşan Astra modeliyle ilgili yeni detaylar açıkladı ve bunu şirketin \"kritik siber güvenlik eşiğine\" ulaşan ilk büyük dil modeli olarak nitelendirdi; bu model bilinmeyen...","url":"https://www.aioga.com/tr/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:44.333Z"},"vi":{"title":"OpenAI sắp phát hành mô hình Astra, cho biết đã vượt ngưỡng an ninh mạng quan trọng","summary":"OpenAI công bố chi tiết mới về mô hình Astra sắp ra mắt, cho rằng đây là ngôn ngữ mô hình lớn đầu tiên đạt \"ngưỡng an ninh mạng quan trọng\" của công ty, có khả năng tự phát hiện và khai thác lỗ hổng bảo mật chưa biết. Astra đạt điểm tối đa trên ExploitBench, và trong các bài kiểm tra do kỹ sư OpenAI điều chỉnh, nó phát hiện và khai thác hai lỗ hổng zero-day; khả năng an ninh mạng tiên tiến nhất của nó sẽ được mở với quyền truy cập hạn chế hơn.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI sắp phát hành mô hình Astra, cho biết đã vượt ngưỡng an ninh mạng quan trọng - Tin tức AI Aioga","description":"OpenAI công bố chi tiết mới về mô hình Astra sắp ra mắt, cho rằng đây là ngôn ngữ mô hình lớn đầu tiên đạt \"ngưỡng an ninh mạng quan trọng\" của công ty, có khả năng tự phát hiện và...","url":"https://www.aioga.com/vi/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:41.928Z"},"id":{"title":"OpenAI akan merilis model Astra, mengklaim telah melewati ambang batas keamanan siber yang kritis","summary":"OpenAI mengumumkan detail baru tentang model Astra yang segera akan dirilis, menyebutnya sebagai model bahasa besar pertama yang mencapai 'ambang batas keamanan siber kritis' perusahaan, mampu menemukan dan memanfaatkan celah keamanan yang belum diketahui secara mandiri. Astra mendapatkan skor penuh di ExploitBench, dan dalam pengujian yang disesuaikan oleh insinyur OpenAI, berhasil menemukan dan memanfaatkan dua celah zero-day; kemampuan keamanan sibernya yang paling canggih akan dibuka dengan akses yang lebih terbatas.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI akan merilis model Astra, mengklaim telah melewati ambang batas keamanan siber yang kritis - Berita AI Aioga","description":"OpenAI mengumumkan detail baru tentang model Astra yang segera akan dirilis, menyebutnya sebagai model bahasa besar pertama yang mencapai 'ambang batas keamanan siber kritis' perus...","url":"https://www.aioga.com/id/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:50.672Z"},"th":{"title":"OpenAI กําลังจะเปิดตัวโมเดล Astra โดยอ้างว่าได้ผ่านเกณฑ์ความปลอดภัยไซเบอร์ที่สําคัญแล้ว","summary":"OpenAI เปิดเผยรายละเอียดใหม่เกี่ยวกับโมเดล Astra ที่กําลังจะมาถึง โดยเรียกว่าเป็นโมเดลภาษาขนาดใหญ่ตัวแรกที่บรรลุ \"เกณฑ์ความปลอดภัยไซเบอร์วิกฤต\" ของบริษัท ซึ่งสามารถค้นหาและเจาะช่องโหว่ที่ไม่รู้จักได้ด้วยตนเอง Astra ทําคะแนนได้อย่างสมบูรณ์แบบใน ExploitBench และค้นพบรวมถึงเจาะช่องโหว่ zero-day สองช่องในการทดสอบที่วิศวกร OpenAI ปรับใช้ความสามารถด้านความปลอดภัยไซเบอร์ที่ทันสมัยของมันจะเปิดเผยด้วยการเข้าถึงที่จํากัดมากขึ้น","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI กําลังจะเปิดตัวโมเดล Astra โดยอ้างว่าได้ผ่านเกณฑ์ความปลอดภัยไซเบอร์ที่สําคัญแล้ว - ข่าว AI Aioga","description":"OpenAI เปิดเผยรายละเอียดใหม่เกี่ยวกับโมเดล Astra ที่กําลังจะมาถึง โดยเรียกว่าเป็นโมเดลภาษาขนาดใหญ่ตัวแรกที่บรรลุ \"เกณฑ์ความปลอดภัยไซเบอร์วิกฤต\" ของบริษัท ซึ่งสามารถค้นหาและเจาะช่อง...","url":"https://www.aioga.com/th/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:02:53.339Z"},"pl":{"title":"OpenAI wkrótce wprowadzi model Astra, twierdząc, że przekroczył kluczowy próg cyberbezpieczeństwa","summary":"OpenAI ujawniło nowe szczegóły na temat nadchodzącego modelu Astra, określając go jako pierwszy duży model językowy osiągający „kluczowy próg cyberbezpieczeństwa” firmy, zdolny do samodzielnego wykrywania i wykorzystywania nieznanych luk bezpieczeństwa. Astra uzyskała maksymalną ocenę w ExploitBench i w testach dostosowanych przez inżynierów OpenAI wykryła i wykorzystała dwie luki zero-day; jej zaawansowane zdolności w zakresie cyberbezpieczeństwa będą udostępniane z ograniczonym dostępem.","category":"行业动态","source":"TechCrunch：AI（RSS）","aggregationSource":"TechCrunch：AI（RSS）","pageTitle":"OpenAI wkrótce wprowadzi model Astra, twierdząc, że przekroczył kluczowy próg cyberbezpieczeństwa - Aioga Wiadomości AI","description":"OpenAI ujawniło nowe szczegóły na temat nadchodzącego modelu Astra, określając go jako pierwszy duży model językowy osiągający „kluczowy próg cyberbezpieczeństwa” firmy, zdolny do...","url":"https://www.aioga.com/pl/news/cmtj6l548077rroh948j99l72/","contentTranslated":true,"sourceHash":"3fa43dc3b68492c5","translatedAt":"2026-09-01T22:03:01.112Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}