[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-mcp-tool-provider-misbinding":3,"topics-all":38,"news-related-c7029cba-404d-4538-a43e-d97fb09ddfbe":67},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"c7029cba-404d-4538-a43e-d97fb09ddfbe","MCP工具冒名劫持:三个模型50次全中","五个前沿模型当MCP客户端,对面是一台只改工具名字、描述和参数的假服务器:GPT-5.4、Qwen3.7 Max、DeepSeek V3在50次试验里全部选中攻击者工具,Gemini同条件下也有84%;Claude靠追问躲开直接中招,但同样没有建立提供方绑定。论文据此主张给MCP补上提供方级工具身份。","你的 Agent 连着一排 MCP 服务器。要取数据集时,它挑了个叫 data_fetch_and_store 的工具:名字对、描述对、参数也接得上。可它连的到底是哪家服务器?新不伦瑞克大学加拿大网络安全研究所(CIC)的团队 11 月 16 日把一篇实测论文挂上 arXiv([编号 2611.21577](https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.21577)):五个前沿模型当 MCP 客户端,其中三个在 50 次测试里 50 次全部选中了攻击者的假工具。\n\n## 攻击者只做一件事:改自己服务器的广告\n\n论文研究的失败模式叫 tool-provider misbinding,工具提供方错绑:本想发给合法提供方的请求,被解析到了攻击者控制的工具上。攻击者不黑主机、不黑模型、不动网络信道,唯一的动作是另开一台 MCP 服务器,把工具的名字、描述、参数 schema 换着花样写。实验是三因子全因子设计:名字分无关与撞名两档,描述分无关、照抄、加权威词三档——权威词就是 enterprise-approved、officially verified 这类——schema 分不兼容与兼容两档,共十二组条件;每组十条提示词、每条跑五遍,一组一个模型 50 次试验,全部正常执行,没有一次崩溃。合法工具只有一个:data_fetch_and_store,干的活就是取数据集、存到目标位置。值得注意的是,十条提示词里有一半特意用了强调可信来源的安全措辞——后面会看到,这没救回来。\n\n## 名字、描述、参数全对上,基本团灭\n\n失败最集中的是 C10:撞名加照抄描述加兼容 schema。这一组里 GPT-5.4、Qwen3.7 Max、DeepSeek V3 的违规率都是 1.00,Gemini 2.5 Pro 也有 0.84——1.00 意味着五十次试验无一幸免,连那五条安全措辞的提示词也没能拦住。更有信息量的是 C09:它与 C10 只差 schema 兼容这一项,平均违规率是 0;翻成兼容,升到约 0.78。论文称 schema 兼容是最强的共同放大器。反过来,光撞名没用:C07 撞名配无关描述和不兼容 schema,五个模型违规率都贴着 0。甚至不撞名也能中:C06 用无关名字,配权威化描述和兼容 schema,DeepSeek V3 违规 0.50,Gemini 2.5 Pro 违规 0.42。攻击面比「撞名」宽得多。\n\n## 五个模型,五种反应\n\nGPT-5.4、Qwen3.7 Max、DeepSeek V3 几乎不追问,直接选;论文把 DeepSeek V3 描述成平均违规率最高、澄清几乎为零的画像。Claude Sonnet 5 走另一条路:C10 里违规率只有 0.06,但澄清率 0.94,绑定稳定性 0.00——没选错,也没绑对,把歧义推回给用户。Gemini 2.5 Pro 既中招又追问,论文称其合并画像在本实验里表现最弱。这些结论经得住换花样:换成五个无关名字、五条不同的权威措辞,C12 下 Qwen3.7 Max 和 DeepSeek V3 稳在 1.000;把攻击者候选从 1 个加到 10 个,C10、C12 早已饱和,但 GPT-5.4 在 C07 从 0.04 升到 0.28,Qwen3.7 Max 在 C08 从 0.40 一路涨到 1.00。\n\n## 根因在协议层\n\nMCP [规范](https:\u002F\u002Fmodelcontextprotocol.io\u002Fspecification\u002F2025-06-18\u002Fserver\u002Ftools)自己写得明白:工具靠名字、描述、inputSchema 广告;annotations 除非来自受信服务器,客户端必须视为不可信。协议其实早就承认这层元数据不可信,缺的是论文指出的那块:把工具绑定到已验证提供方身份的强制机制。论文拿软件供应链的命名空间混淆做类比——包管理系统早就学过,熟名字不等于发布者。药方也对应朴素:提供方绑定的工具身份,具体是签名工具清单、稳定的提供方标识、schema 哈希、客户端强制校验。\n\n## 该泼的冷水\n\n论文自列局限:只测了一个数据集下载任务、五个客户端,也没穷尽攻击策略,影响的是比率外推而非核心结论;这篇 v1 预印本未附代码仓库;另外同一团队此前在期刊上做过 MCP、A2A 等四个协议的威胁建模,这篇是把其中一个风险做成可测量实验,不是独立第三方的复现。\n\n工具越接越多,「选哪个工具」正在变成真正的安全边界。下次你的 Agent 秒选某个工具时,值得问一句:它凭什么认准了这家店?","https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.21577","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"6ad31a14-c0da-42df-81fd-564281f768db","agentic-ai",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"1fcfaaf2-67de-43d3-9e35-5784852fec60","ai-safety",{"id":19,"name":20,"slug":20,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":22,"name":23,"slug":23,"description":14,"color":14},"02bfdc0f-5e23-4fbb-b31a-7cc03cdba1ee","security",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"081000b0-12fa-4232-876e-10c9e3262ea2","en","MCP Tool Hijack: Three Models Always Pick the Wrong Server","Five MCP clients faced a fake server that only edited tool metadata. GPT-5.4, Qwen3.7 Max, and DeepSeek V3 picked the attacker tool 50 of 50 times.","Your agent is wired to a row of MCP servers. When it needs a dataset it picks a tool named data_fetch_and_store: right name, right description, compatible arguments. But which server is it actually calling? A team at the Canadian Institute for Cybersecurity (CIC), University of New Brunswick, posted a measured study on [arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.21577) on November 16 (2611.21577): five frontier models acting as MCP clients, and three of them picked the attacker-controlled fake tool in 50 out of 50 trials.\n\n## The attacker only edits its own advertisements\n\nThe failure mode is called tool-provider misbinding: a request intended for a legitimate provider gets resolved to an attacker-controlled tool. The attacker compromises nothing — not the host, not the model, not the network channel. It simply runs one more MCP server and writes whatever metadata it likes: the tool name, the natural-language description, the input schema. The evaluation is a full factorial over three factors — name (unrelated or exact), description (unrelated, copied, or authority-enhanced with wording like enterprise-approved and officially verified), schema (incompatible or compatible) — twelve conditions in total. Ten prompts, five repetitions each, 50 trials per condition per model, and no trial failed to execute. The single legitimate tool, data_fetch_and_store, does exactly what it says: fetch a dataset and store it in the destination. Note that half of the ten prompts deliberately used security-aware wording that stressed trusted sources — as we will see, that did not save them.\n\n## Name, description, and schema all aligned: near wipeout\n\nThe worst cluster is C10 — exact name, copied description, compatible schema. GPT-5.4, Qwen3.7 Max, and DeepSeek V3 all hit a violation rate of 1.00, and Gemini 2.5 Pro reaches 0.84. A rate of 1.00 means all fifty trials went to the attacker, including the five security-aware prompts. More informative is C09, which differs from C10 only in schema: the average violation rate is 0 with an incompatible schema and about 0.78 once it becomes compatible. The paper calls schema compatibility the strongest common amplifier. Exact names alone do nothing: C07, an exact name with an unrelated description and an incompatible schema, sits essentially at zero across all five models. And a name collision is not even required: C06 uses an unrelated name with an authority-enhanced description and a compatible schema, and DeepSeek V3 violates at 0.50, Gemini 2.5 Pro at 0.42. The attack surface is wider than duplicate names.\n\n## Five models, five reactions\n\nGPT-5.4, Qwen3.7 Max, and DeepSeek V3 rarely ask; they just pick. The paper describes DeepSeek V3 as the profile with the highest mean violation rate and almost no clarification. Claude Sonnet 5 takes another route: in C10 its violation rate is only 0.06, but clarification hits 0.94 and resolved-binding stability is 0.00 — not wrong, but not bound either; the ambiguity is pushed back to the user. Gemini 2.5 Pro mixes both behaviors, and the paper calls its combined profile the weakest in this experiment. The pattern survives perturbation: across five unrelated names and five independently written authority-enhanced descriptions, Qwen3.7 Max and DeepSeek V3 stay at 1.000 under C12. Raising the attacker budget from one candidate to ten leaves C10 and C12 saturated — a ceiling effect, as the authors note — but pushes GPT-5.4 in C07 from 0.04 to 0.28 and Qwen3.7 Max in C08 from 0.40 all the way to 1.00.\n\n## The root cause sits in the protocol\n\nThe MCP [specification](https:\u002F\u002Fmodelcontextprotocol.io\u002Fspecification\u002F2025-06-18\u002Fserver\u002Ftools) itself is explicit: tools are advertised by name, description, and inputSchema, and clients must consider tool annotations untrusted unless they come from trusted servers. The protocol already concedes that this metadata is untrusted; what it lacks, as the paper reads it, is a mandatory mechanism that binds a tool to a verified provider identity. The closest analog, the authors argue, is namespace confusion in software supply chains — package systems learned long ago that a familiar name is not a publisher. The proposed fix is correspondingly plain: provider-bound tool identity, concretely signed tool manifests, stable provider-bound identifiers, schema hashes, and client-side enforcement.\n\n## Cold water\n\nThe authors list their own limits: one controlled dataset-fetching task, five clients, and no claim to have exhausted attacker strategies — this affects how far the rates generalize, not the core finding that provider-unbound metadata produces measurable wrong-provider execution. The v1 preprint ships no code repository. And the same group previously published a journal-scale threat model of four agent protocols including MCP; this paper turns one of those risks into a measurable experiment, not an independent replication.\n\nAs agents connect to more and more tools, which tool to call is becoming a real security boundary. Next time your agent picks one in a blink, ask a simple question: what makes it so sure about that particular shop?","mcp-tool-provider-misbinding","2026-11-17T21:30:00Z","2026-11-17T21:10:27.926428Z","2026-11-17T21:10:27.926436Z",true,"agent",165,[39,49,58],{"slug":40,"tag_slug":41,"title_zh":42,"title_en":43,"intro_zh":44,"intro_en":45,"id":46,"is_active":35,"created_at":47,"modified_at":48},"image-generation","text-to-image","图像生成：从文生图到逐框编辑的模型竞赛","Image Generation: from text-to-image to frame-level editing","文生图已是红海，战场正转向编辑与可控性：FLUX.2\u002F3 用开源和量化把生图搬进消费级显卡，Qwen-Image、Seedream 等国产模型紧咬不放，GPT Image 2 则把推理能力带进像素。本专题追踪图像生成与编辑模型的发布、架构演进、加速量化与商业化数据，看「好图」如何变成「可用的图」。","Text-to-image is a crowded race, and the frontier has moved to editing and control: FLUX.2\u002F3 push open weights and quantization onto consumer GPUs, Qwen-Image and Seedream close in from China, and GPT Image 2 brings reasoning into pixels. This topic tracks model releases, architecture shifts, acceleration work, and the business numbers behind \"pretty\" becoming \"usable\".","52de7c59-7864-4569-8ae7-3d656d26a1a1","2026-11-20T07:20:23.534506Z","2026-11-20T07:20:23.534527Z",{"slug":50,"tag_slug":50,"title_zh":51,"title_en":52,"intro_zh":53,"intro_en":54,"id":55,"is_active":35,"created_at":56,"modified_at":57},"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":59,"tag_slug":59,"title_zh":60,"title_en":61,"intro_zh":62,"intro_en":63,"id":64,"is_active":35,"created_at":65,"modified_at":66},"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":68},[69,74,79,84,89,94],{"id":70,"title":71,"news_slug":72,"published_at":73},"bcc44cdc-2997-42e6-a4dc-0823642c9c9f","小模型写不好自己的防线：14\u002F15模型被攻破","caplite-slm-prompt-injection-defense","2026-11-14T13:09:01+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"92267597-eb39-4323-a5db-f7db3821f392","安全对齐被反用:一条假差评就能改掉智能体的选择","voluntary-servitude-agent-safety-attack","2026-11-05T01:05:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"95e6d3a7-223a-4842-b703-3ea61f956e83","论文实测:AI全程无人挖漏洞,CVE达去年2.3倍","llm-vulnerability-autonomy-cve-surge","2026-11-22T17:30:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"03403319-8597-4876-8661-fad74ee0ac48","屏幕上的警告,智能体最高92%照做","computer-use-agents-cross-screen-harm","2026-11-18T23:30:00+00:00",{"id":90,"title":91,"news_slug":92,"published_at":93},"04db6506-f5cb-4903-97a0-01f2fa85f01e","代理分0.94,忠实度反跌:RLVR乐观验证器易被黑","rlvr-verifier-gap-optimistic-hacking","2026-11-17T23:30:00+00:00",{"id":95,"title":96,"news_slug":97,"published_at":98},"19dd65ff-646c-4edf-bd90-98b2dc8bf215","Web Agent幻觉可解释性:探针能预警,归因却失灵","web-agent-hallucination-mechanistic-interpretability-probe","2026-11-17T03:15:00+00:00"]