内容来源：https://www.engadget.com/ai/openai-will-amend-defense-department-deal-to-prevent-mass-surveillance-in-the-us-050637400.html?src=rss

内容总结：

美国人工智能公司OpenAI首席执行官萨姆·阿尔特曼近日表示，公司将修改与美国国防部（亦称战争部）的合作协议，明确禁止其AI系统用于针对美国公民的大规模监控。阿尔特曼在社交平台X上发布了一份此前发送给员工的内部备忘录，称将在协议中增加条款，以特别强调禁止将AI技术用于国内监控。

备忘录指出，根据美国宪法第四修正案、《1947年国家安全法》及《1978年外国情报监视法》等适用法律，该AI系统不得被故意用于对美国公民及国民进行国内监控。国防部已明确理解此限制条款禁止对美国公民或国民进行有意追踪、监视或监控，包括通过采购或使用商业获取的个人身份信息。

阿尔特曼在备忘录中称，国防部已确认其情报机构（包括国家安全局）未获合同修改前不会使用OpenAI服务。他同时强调，若接到其认为违宪的命令，宁愿入狱也不会执行。

阿尔特曼承认，公司不应仓促于2026年2月27日（周五）推进此项合作，因相关问题“极其复杂且需要清晰沟通”。他解释称，公司当时试图“缓和事态并避免更糟结果”，但最终显得“机会主义”。据悉，OpenAI宣布与国防部合作前，特朗普总统刚下令所有美国政府机构停止使用Anthropic公司的Claude及其他服务。

此前，国防部及部长皮特·赫格塞斯曾向Anthropic施压，要求移除其AI系统的安全限制，以用于包括大规模监控和全自主武器开发在内的所有“合法”用途。Anthropic拒绝妥协，并声明“任何恐吓或惩罚”都不会改变其“反对国内大规模监控及全自主武器的立场”，特朗普随后签署了上述禁令。国防部还曾启动将Anthropic列为“供应链风险”的程序，此类认定通常针对被认为与本国政府合作的中国企业。

阿尔特曼透露，在与美官员沟通中，他重申不应将Anthropic列为供应链风险，并希望国防部能向其提供与OpenAI相同的合作条款。他在周末的X平台问答中澄清，虽不了解Anthropic协议细节，但若条款相同，他认为Anthropic应接受该协议。

OpenAI与国防部合作消息曝光后，Anthropic的Claude应用在苹果应用商店免费榜排名升至第一，超过ChatGPT和谷歌Gemini。Anthropic趁势推出记忆导入工具，方便用户从其他聊天机器人切换至Claude。据Sensor Tower数据，ChatGPT的日卸载量在此期间激增295%。

中文翻译：

OpenAI将修订与美国国防部的合作协议，明确禁止利用其技术在美国境内实施大规模监控。公司首席执行官萨姆·阿尔特曼表示，将在协议中增加条款，明文禁止任何使用OpenAI系统监视美国公民的行为。阿尔特曼在社交媒体X上发布了一份此前发送给员工的内部备忘录，称公司将调整协议内容，通过补充具体条款使这一立场更加清晰。备忘录中特别指出：

"依据包括美国宪法第四修正案、1947年《国家安全法》、1978年《外国情报监视法》在内的相关法律，本人工智能系统不得被蓄意用于对美国公民及国民进行境内监控。
为避免歧义，国防部理解此限制条款禁止通过采购或使用商业渠道获取的个人身份信息，对美国公民或国民实施有意的追踪、监视或监控行为。"

阿尔特曼在备忘录中强调，国防部已确认未经合同修订，其下属情报机构（包括国家安全局）不得使用OpenAI服务。他同时表示，若接到其认为违宪的命令，宁可承担法律责任也绝不会执行。

这位首席执行官还在备忘录中承认，公司本不应仓促于2026年2月27日（周五）公布该协议，因为相关议题"极其复杂且需要清晰的沟通"。他解释称，公司当时"试图缓和事态并避免更糟糕的后果"，但最终举措"显得过于投机"。值得回顾的是，OpenAI宣布此项合作的时间点，恰在前总统特朗普下令所有美国政府机构停止使用Claude及其他Anthropic服务之后。需要说明的是，Anthropic公司自2024年起开始与美国政府开展合作。

此前，美国国防部及部长皮特·赫格塞斯曾持续向Anthropic施压，要求其解除人工智能系统的安全限制，以允许该技术用于所有"合法"用途，包括大规模监控和全自主武器研发。Anthropic坚决拒绝了赫格塞斯的要求，并发表声明称"任何恐吓或惩罚都无法改变公司对境内大规模监控及全自主武器的立场"，此举直接导致了特朗普发布行政禁令。国防部还曾启动将Anthropic列为"供应链风险"的审查程序——此类认定通常针对被认为与其本国政府存在合作的中资企业。

阿尔特曼透露，在与美国官员的会谈中，他多次重申不应将Anthropic列为供应链风险，并希望国防部能向其提供与OpenAI相同的合作条款。在周末于X平台举行的问答活动中，阿尔特曼澄清自己并不了解Anthropic协议的具体内容及其与OpenAI协议的差异，但他认为若条款相同，Anthropic理应接受该协议。

OpenAI合作消息曝光后，Anthropic开发的应用程序迅速登顶苹果应用商店免费榜，超越ChatGPT和谷歌Gemini。Anthropic趁势推出记忆导入工具，方便用户从其他聊天机器人平台迁移至Claude。据Sensor Tower数据显示，同期ChatGPT应用的日卸载量激增295%。

以下为内部公告重发内容：
我们正与国防部协作修订协议，通过补充条款明确我们的核心原则：

1. 除现有条款外，我们将在协议中增补如下内容：
   "• 依据包括美国宪法第四修正案……" ——萨姆·阿尔特曼 (@sama) 2026年3月3日

英文来源：

OpenAI will amend Defense Department deal to prevent mass surveillance in the US
Sam Altman said the company will add language to explicitly prohibit the use of OpenAI systems to spy on Americans.
OpenAI’s Sam Altman said the company will amend its deal with the Defense Department (or the Department of War) to explicitly prohibit the use of its AI system on mass surveillance against Americans. Altman has published an internal memo previously sent to employees on X, telling them that the company will tweak the agreement to add language to make that point especially clear. Specifically, it says:
“Consistent with applicable laws, including the Fourth Amendment to the United States Constitution, National Security Act of 1947, FISA Act of 1978, the AI system shall not be intentionally used for domestic surveillance of U.S. persons and nationals.
For the avoidance of doubt, the Department understands this limitation to prohibit deliberate tracking, surveillance, or monitoring of U.S. persons or nationals, including through the procurement or use of commercially acquired personal or identifiable information.”
Altman has also claimed in the memo that the agency affirmed that its services will not be used by its intelligence agencies, including the NSA, without a modification to their contract. He added that if he received what he believed was an unconstitutional order, he would rather go to jail than follow it.
In addition, the OpenAI CEO has admitted in the memo that the company shouldn’t have rushed to get the deal out on Friday, February 27, since the issues were “super complex and demand clear communication.” Altman explained that the company was “trying to de-escalate things and avoid a much worse outcome” but it “looked opportunistic” in the end. If you’ll recall, OpenAI announced the partnership shortly after President Trump ordered all US government agencies to stop using Claude and any other Anthropic services. To note, Anthropic started working with the US government in 2024.
The Defense Department and Secretary Pete Hegseth had been pressuring Anthropic with to remove its AI’s guardrails so that it can be used for all “lawful” purposes. Those include mass surveillance and the development of fully autonomous weapons. Anthropic refused to bow down to Hegseth’s demands and in a statement said that “no amount of intimidation or punishment” will change its “position on mass domestic surveillance or fully autonomous weapons.” Trump issued the order as a result. The Defense Department had also taken the first steps to designate Anthropic as a “supply chain risk,” which is typically reserved for Chinese companies believed to be working with their country’s government.
Altman said that in his conversations with US officials, he reiterated that Anthropic shouldn’t be designated as a supply chain risk and that he hoped the Defense Department would offer it the same deal OpenAI agreed to. In an AMA session on X over the weekend, Altman clarified that he didn’t know the details of Anthropic’s agreement and how it differed from the one OpenAI signed. But if it had been the same, he thought Anthropic should have agreed to it.
After the news broke out about OpenAI’s deal, Anthropic climbed its way to the number one spot of the App Store's Top Free Apps leaderboard, beating out both ChatGPT and Google Gemini. Anthropic, capitalizing on Claude’s sudden popularity, launched a memory import tool to make switching to its chatbot from another company’s easier. Meanwhile, uninstalls for ChatGPT’s jumped by 295 percent day-over-day, according to Sensor Tower.
Here is re-post of an internal post:
We have been working with the DoW to make some additions in our agreement to make our principles very clear.

1. We are going to amend our deal to add this language, in addition to everything else:
   "• Consistent with applicable laws,…— Sam Altman (@sama) March 3, 2026