Systems Research系统研究Oct 25, 20262026年10月25日Note短评

In public-sector AI, the contract decides access, and capability comes second.在公共部门的 AI 采购里，决定谁能进场的是合同，能力只排第二。

A bright, sunlit government checkpoint under a clear sky. A plain, dull grey faceless figure is waved through an open turnstile glowing warm gold because it carries one oversized folder sealed with a big gold wax stamp; beside it, two much taller, ornate figures sculpted from glass and gold filigree — clearly more powerful — stand empty-handed behind a lowered barrier, shut out. White caption: Smarter, still banned. A Chinese line sits below it: 更聪明，却被挡在门外.阳光下晴空中的一处政府安检口。一个普通、灰暗、没有面孔的人形因为手里拿着一个盖了巨大金色蜡印的大文件夹，被笑着放行通过一道泛着暖金色光的闸机；它旁边，两个高大得多、由玻璃与金色花丝雕成的华丽人形——明显更强大——空着手被拦在一道落下的横杆后面，挡在门外。白色大字写着：Smarter, still banned，下面一行中文：更聪明，却被挡在门外。Image: WAKENEX图片：WAKENEX

Sweden proved it on October 24. Its Government Offices banned two consumer AI tools, ChatGPT and Claude, from staff computers and kept just one, Microsoft Copilot — and the block had nothing to do with which model is smarter. Copilot even runs Claude: Microsoft began offering Anthropic's models inside it this year, disabled by default in the EU. So the same weights a government bans can still reach it through an approved vendor.

What cleared Copilot was a contract and a security review Microsoft agreed to harden. ChatGPT and Claude were never put through that review. The Government Offices' director of digitalisation, Mikael Wettercrantz, said it plainly: using AI without a formal agreement has "consequences for our information security," because the agreement sets what the provider can and cannot do with a civil servant's inputs. A consumer chatbot ships that query to a server where a foreign government can later compel it; a contracted service names who is liable when it leaks.

So "best model" and "approved vendor" are two different games, and people keep conflating them. A lab can top every benchmark and still be shut out of an entire government, while its model quietly runs inside the tool that got in.

The honest counter: Copilot's tenant isolation and admin controls are genuine security engineering, and the review was substantive. Fair. But Sweden's stated basis was the missing agreement, and it never measured the others against it.

If you build on AI, the lesson is blunt: for regulated buyers your moat is the data-processing agreement and the jurisdiction you can run in. Capability gets you demos; contracts get you deployed.

This breaks if a government procurement openly picks the top-benchmark model over a contracted but weaker vendor.

瑞典在 10 月 24 日把这一点演示了出来。它的政府办公厅（Government Offices）把 ChatGPT 和 Claude 这两个面向大众的 AI 工具从工作电脑上封禁，只保留了微软 Copilot 这一个——而这次封禁和哪个模型更聪明毫无关系。Copilot 里其实就跑着 Claude：微软今年已经开始在它里面提供 Anthropic 的模型（在欧盟默认关闭）。所以同一套被政府封禁的模型权重，照样能通过一个被批准的供应商送进这个政府。

让 Copilot 过关的，是一份合同，以及一次微软同意加固的安全审查。ChatGPT 和 Claude 根本没有被拿去做这项审查。政府办公厅数字化负责人 Mikael Wettercrantz 把话说得很直白：使用没有正式协议的 AI，会对“我们的信息安全”造成影响，因为这份协议规定了供应商可以拿公务员输入的内容做什么、不能做什么。一个面向大众的聊天机器人，会把这条查询发到某台服务器上，而外国政府日后可以依法强制调取；有合同约束的服务，则写明了数据泄露时由谁担责。

所以“最强的模型”和“被批准的供应商”，是两场不同的游戏，而人们总把它们混为一谈。一家实验室可以在每一项跑分上登顶，却照样被整个政府拒之门外——与此同时，它的模型正悄悄跑在那个被放行的工具里。

诚实的反方意见是：Copilot 的租户隔离和管理控制是实打实的安全工程，这次审查也是有实质内容的。这没错。但瑞典给出的理由是对方没有协议，而且它从没把其他几家拿来和这个标准比对。

如果你在 AI 之上做产品，这个教训很干脆：面向受监管的买家，你的护城河是那份数据处理协议，以及你能在他们辖区内运行的能力。能力只能帮你拿到演示机会；合同才能让你真正被部署。

如果出现一笔政府采购，公开选择跑分最高的模型，而放弃一个有合同但更弱的供应商，那这个判断就不成立。

WAKENEX / NOTESWAKENEX / 短评

Add WAKENEX as a preferred source on Google在 Google 把 WAKENEX 设为首选来源

READ NEXT接着读

OpenAI halved Terra's price, and the discount moved the cost onto you.OpenAI 把 GPT-6.1 Terra 的价格砍了一半，而这笔折扣悄悄把一项成本转移到了你身上。

Mochatrade sells Indian retail leveraged price exposure to US stocks.Mochatrade 卖给印度散户的，是对美股价格的带杠杆敞口。

NetSuite now lets you swap the AI model, not the permission gate.甲骨文旗下 NetSuite 的新说辞——“自带 AI”——是一道挪了位置的护城河，而不是一堵倒下的墙。