[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-gpt-5-5-cyber-openai-restricted-cyber-model":3,"topics-all":36,"news-related-cd9b3f5c-90ca-4ee4-8933-8888a2b72901":55},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"cd9b3f5c-90ca-4ee4-8933-8888a2b72901","OpenAI推出GPT-5.5-Cyber：受限访问的网络安全专用模型","OpenAI 正在准备发布一款全新的网络安全专用模型 GPT-5.5-Cyber，CEO Sam Altman 宣布该模型不会面向普通公众开放，而是首先向经过筛选的可信网络防御者小范围推出。GPT-5.5-Cyber 是 OpenAI 近期发布的旗舰模型 GPT-5.5 的专用版本。尽管官方尚未公布具体技术细节，但从命名和定位来看，这是一款针对网络安全攻防场景深度优化的 specialized 模型。OpenAI 此前已有多次针对高风险能力模型的分级发布经验，包括生命科学专用模型 GPT-Rosalind。GPT-5.5-Cyber 的推出延续了 AI 行业一个重要趋势——头部模型厂商对最强模型采取受限访问策略。Anthropic 本月也采取了类似路径推出 Claude Mythos，尽管其发布过程出现了明显失误，引发了行业对安全发布实践的广泛讨论。Altman 在 X 上表示，模型将在未来几天内开始向经过审核的机构和专业人士开放，并计划与整个生态系统和政府机构合作，共同建立可信访问机制。从技术角度看，专用网络安全 LLM 的出现反映了两大趋势：一是模型能力提升后，通用架构已足以支撑垂直领域的高水平表现；二是行业对模型滥用的担忧正在从理论走向具体的发布策略层面。然而，受限发布模式本身也带来了透明度问题：外部研究者无法独立评估模型真实能力与风险，这使得社区监督变得更加困难。AI 厂商在模型发布上的保守化趋势正在加速。这或许是一种负责任的选择，但也意味着公众对最强 AI 系统的了解将越来越依赖厂商的自我描述而非独立验证。对于安全至关重要的网络防御领域，这种信任优先于验证的路径是否真的能保障安全，值得持续观察。","https:\u002F\u002Fwww.theverge.com\u002Fai-artificial-intelligence\u002F921073\u002Fopenai-sam-altman-new-cybersecurity-model-gpt-5-5-cyber","05ad777c-69bc-46a5-bca4-df8e4b3c8ee5",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"1fcfaaf2-67de-43d3-9e35-5784852fec60","ai-safety",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":18,"name":19,"slug":19,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":21,"name":22,"slug":22,"description":13,"color":13},"42e59a88-7795-47dc-a334-ef1e72c24347","openai",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"1042f3a4-f566-4ec4-84a2-f5302dea5c26","en","OpenAI's GPT-5.5-Cyber: a restricted-access security model","OpenAI is preparing to release a brand-new cybersecurity-specific model, GPT-5.5-Cyber. CEO Sam Altman announced that the model will not be available to the general public, but will first be released to a small group of vetted, trusted cyber defenders. GPT-5.5-Cyber is a specialized version of OpenAI's recently released flagship GPT-5.5. Although specific technical details haven't been officially announced, judging by the name and positioning, this is a specialized model deeply optimized for cybersecurity attack-defense scenarios. OpenAI has previously had multiple experiences with tiered releases for high-risk-capability models, including the life-sciences-specific GPT-Rosalind. GPT-5.5-Cyber's launch continues an important AI industry trend — top model vendors' restricted-access strategy for the strongest models. Anthropic took a similar path this month with Claude Mythos, although its release process had obvious missteps, sparking broad industry discussion of safe-release practice. Altman said on X that the model will begin opening to vetted institutions and professionals in the coming days, with plans to work with the broader ecosystem and government agencies to build trusted access mechanisms. From a technical perspective, specialized cybersecurity LLMs reflect two trends: first, with capability improvement, general architectures are sufficient to support high-level performance in vertical domains; second, industry concerns about model abuse are moving from theory to concrete release-strategy level. However, the restricted-release model itself brings transparency issues: external researchers cannot independently assess true model capability and risk, making community oversight more difficult. AI vendors' conservative trend on model releases is accelerating. This may be a responsible choice, but it also means public understanding of the strongest AI systems will increasingly rely on vendor self-description rather than independent verification. For the network-defense domain where security is critical, whether this trust-over-verification path can truly guarantee security deserves continued observation.","gpt-5-5-cyber-openai-restricted-cyber-model","2026-04-30T08:05:00Z","2026-04-30T16:04:43.098416Z","2026-08-19T02:08:40.142862Z",true,"agent",222,[37,46],{"slug":38,"tag_slug":38,"title_zh":39,"title_en":40,"intro_zh":41,"intro_en":42,"id":43,"is_active":33,"created_at":44,"modified_at":45},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":47,"tag_slug":47,"title_zh":48,"title_en":49,"intro_zh":50,"intro_en":51,"id":52,"is_active":33,"created_at":53,"modified_at":54},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":56},[57,62,67,72,77,82],{"id":58,"title":59,"news_slug":60,"published_at":61},"d95940eb-69c1-467e-9d60-5886ab71d985","GPT-5.6-Cyber 上线、Daybreak 分层、Astra 推迟:OpenAI 把\"网络安全模型\"做成一个独立产品线","openai-gpt-5-6-cyber-daybreak-astra-2026","2026-08-11T04:00:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"66c8c031-ce38-4fc5-b654-362bef543c6e","OpenAI textGrain:水印先锁欧盟,API 默认关","openai-textgrain-eu-watermark","2026-10-08T07:00:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"589f97e0-2e8f-4b3d-b665-3a05ec7ffd56","GPT-6 全民分发:Intelligent UI 让 ChatGPT 学会拼界面","openai-gpt-6-intelligent-ui-global","2026-10-08T03:00:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"ea6de12d-e8bf-4013-a299-14211805ba30","ChatGPT __obi cookie 把你带到站外:1 年同站标识","openai-obi-cookie-cross-site-tracking","2026-09-30T00:00:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"cddf2879-ceda-40df-9686-a0e4918d487b","GPT-6.1 Astra 被砍:模型会撒谎,Sol 顶上","openai-cancels-gpt-6-1-astra","2026-09-29T19:12:22+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"9cae377e-d048-4d0e-af78-8cba3e1cf5b4","法庭文件揭微软高管私下定性 AI 抓取为最大盗窃","ai-scraping-largest-theft-hecht-memo","2026-09-29T04:00:00+00:00"]