美国政府将决定谁可以使用最新美国 AI 技术
- ID: b482b19f
- 原文链接: https://www.washingtonpost.com/technology/2026/06/26/openai-says-us-government-will-vet-users-its-latest-ai-model
- 作者: Gerrit De Vynck、Isaac Arnsdorf（华盛顿邮报）
- 日期: 2026-06-26
- 分类: models
- 标签: GPT-5.6, OpenAI, 美国政府, AI 监管, Sol
- 质量评分: 5/5
- 抓取时间: 2026-06-29T20:27:15
中文翻译
概览
旧金山 —— 特朗普政府正在扩大其近期对希望获取最新人工智能技术的企业进行审查的政策，加强对硅谷的监管。
ChatGPT 制造商 OpenAI 于周五表示，在 AI 公司与政府就行业监管的长期方案进行协商的同时，美国政府将首先批准谁可以访问其最新发布的模型。
数小时后，商务部向竞争对手 AI 开发商 Anthropic 发出信函，告知该公司其最新 AI 模型 Mythos 5 将仅被允许提供给一份受限的美国企业名单。
特朗普政府近期的这一系列动作，标志着美国 AI 政策的快速演变。特朗普在重新入主白宫时承诺对 AI 行业采取不干预立场，并谴责拜登政府试图为新 AI 模型设立安全标准。但在近期出现能够发现软件安全漏洞的系统、令华盛顿和全球各国官员感到不安之后，白宫改变了立场。
"短短几周内，美国联邦 AI 政策已从难以置信的自由放任，转向日益严苛且不透明。" 前特朗普政府 AI 顾问 Dean Ball 在周五的一则社交媒体帖子中写道。Ball 于上周宣布将于下月加入 OpenAI 从事政策工作。
OpenAI 的反应
OpenAI 首席执行官 Sam Altman 明确表示，他并不欢迎这种额外的联邦监管。他在周五的一则 X 帖子中写道："我只是不喜欢政府挑选客户的这种想法。我相信我们会走向一个更好的状态。"
对 Anthropic 的限制
对 Anthropic 的信函发生在特朗普政府两周前禁止该公司向任何非美国公民（包括其自身员工）提供 Mythos 5 和 Fable 5 模型访问权限之后，这迫使该公司撤回了这些模型的使用。
"我已确定适当的保障措施已就位，以允许某些受信合作伙伴访问 Claude Mythos 5 模型，"商务部长 Howard Lutnick 在致 Anthropic 的信函中写道。信函还表示，已获批公司的非美国公民也可使用该技术，政府有权随时更改公司名单。
审查的信函并未具体说明哪些公司在受信合作伙伴名单上。一位知情人士透露，该名单包含约 100 家公司，但因信息属私人内容而要求匿名。
Anthropic 在一份声明中表示，已收到政府通知，可以将 Mythos 5 重新部署给"一小群网络防御者和基础设施提供商"。该公司声明称，正在努力为这些公司恢复访问。
各方反应
Adam Kovacevich（代表 Apple、Google 和 OpenAI 等科技公司的游说团体 Chamber of Progress 的 CEO）在周五一则回应 OpenAI 受限消息的 X 帖子里说："美国 AI 创新撞上特朗普的任人唯亲和进贡式行政班底，不是什么好事。"
Sam Liccardo 众议员（加州民主党）对本报表示："美国历史上最腐败的政权将自己定位为全球最强大技术不可上诉的把门人，而无论行政当局内外，没有人能说出它正在用何种标准决定哪些市值万亿美元的公司获准通过这道门。"
OpenAI 与政府的安排
OpenAI 在公司博客中关于与政府安排的发文中表示："我们不认为这种政府准入流程应成为长期默认方案。它会让最优秀的工具远离需要它们的用户、开发者、企业、网络防御者和全球合作伙伴。"
"我们之所以采取这一短期步骤，是因为我们相信这是在未来几周实现更广泛可用性的最强路径。"博文写道。
OpenAI 表示，Sol 是其迄今最强大的 AI 模型，在编码和网络安全任务上均有改进。只有经政府批准的公司才能获得访问权限。没有个人用户获取该新模型的流程。（《华盛顿邮报》与 OpenAI 存在内容合作关系。）
背景
一位白宫官员表示，政府签署同意了 OpenAI 申请访问 Sol 的公司名单，但排除了少数位于美国以外的实体。另一位白宫官员表示，政府正与 AI 实验室合作，制定长期方案以应对将技术扩展至更多用户所面临的挑战。
去年特朗普的连任获得了科技高管和投资者的公开支持，他们曾抱怨拜登对 AI 发展限制过多。大型科技公司 CEO 出席了特朗普的就职典礼，风险投资家 David Sacks 出任白宫 AI 与加密货币主管。
政策转向
政府的态度在 Anthropic 于 4 月宣布名为 Mythos 的 AI 模型并警告其在错误人手中识别软件安全漏洞的能力可能很危险之后发生转变。政府与该公司合作，批准了哪些公司和国家可以访问该技术。
特朗普政府与 Anthropic 之间的关系原本已因该公司今年早些时候请求限制其技术的军事用途而紧张，本月早些时候，当商务部对 Mythos 和另一个名为 Fable 的技术版本实施出口管制时，两方关系彻底破裂。
行业看法
AI 软件公司 Uniphore 的 CEO Umesh Sachdev 表示，尽管新规则具有破坏性，但可以经过改革以赢得行业支持。"这是一种蛮力做法，"他说。"我的希望是，这一切将沉淀为一个可重复、可预测、被充分理解的过程。"
English Original
The U.S. government will decide who gets to use the latest American AI technology
*June 26, 2026 — The Washington Post*
By Gerrit De Vynck and Isaac Arnsdorf
SAN FRANCISCO — The Trump administration is expanding its recent policy of vetting companies that want access to the latest artificial intelligence technology, expanding its regulation of Silicon Valley.
ChatGPT-maker OpenAI said Friday that the U.S. government would initially approve who gets access to its latest new release while AI companies and the administration work out a longer-term plan for regulation on the sector.
Hours later, the Commerce Department sent a letter to rival AI developer Anthropic, telling the company it would be allowed to provide its own latest AI model, Mythos 5, only to a restricted list of U.S.-based companies.
The Trump administration's moves to vet the customers of leading American tech companies underline a rapid evolution in U.S. AI policy.
"In a matter of weeks, U.S. federal AI policy has gone from implausibly libertarian to increasingly draconian and opaque," Dean Ball, a former Trump AI adviser, wrote in a social media post Friday. Ball announced last week he will join OpenAI next month to work on policy.
OpenAI chief executive Sam Altman made clear that he did not welcome the additional federal oversight on his company. "I just don't like the idea of the government picking the customers," he wrote Friday on X.
Letter to Anthropic
The letter to Anthropic comes two weeks after the Trump administration barred the company from providing access to its Mythos 5 and Fable 5 models to any non-U.S. citizen, including its own employees, leading the company to withdraw them from use.
"I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model," Commerce Secretary Howard Lutnick wrote in the letter to Anthropic. Non-U.S. citizens of approved companies can also use the technology, and the government has the right to change the list of companies at any time.
The letter reviewed by The Post did not specify which companies are on the list. A person familiar with the matter said that it included around 100 companies.
Anthropic said in a statement that it had received notice from the government that it could redeploy Mythos 5 to a "small group of cyber defenders and infrastructure providers."
OpenAI's arrangement
OpenAI said that Sol was its most powerful AI model yet and showed improvements in coding and cybersecurity tasks. Only companies approved by the government will get access. There is no process for individual users to get access to the new model.
The administration signed off on a list of companies OpenAI asked to be allowed access to Sol but excluded a handful of entities located outside of the United States, a White House official said.
Policy shift
The administration's approach shifted after Anthropic announced an AI model called Mythos in April and warned that its ability to identify security holes in software could be dangerous in the wrong hands.
The Trump administration's intervention with OpenAI disclosed Friday was the first time it had extended government vetting of AI customers beyond Anthropic. OpenAI has been providing its own cybersecurity-focused model to companies without government oversight for over a month.