川普鬆口國防部極可能與 Anthropic「達成 AI 合作協議」，大讚：他們表現漸入佳境

📄完整原文· 由 trafilatura 自動擷取Gemini 翻譯1429 字

The AI giant Anthropic, which was comprehensively banned by the U.S. government, has seen a major turnaround! U.S. President Donald Trump stated on Tuesday that it is "highly likely" that the Department of Defense (DOD) will reach a cooperation agreement with Anthropic. Previously, because Anthropic refused to allow its AI models to be used for "autonomous weapons and mass surveillance," it was labeled a national security risk by the Pentagon, and Trump even ordered federal agencies to ban it entirely. However, with high-level officials from both sides recently holding secret meetings regarding the new "Mythos" model, this legal battle seems to be seeing a breakthrough. (Previous coverage: White House AI Crypto Czar David Sacks steps down and fires back: Anthropic's enterprise coding is great, but they do business by fear-mongering about regulation) (Background supplement: NSA secretly using Anthropic Mythos: The Pentagon's double-dealing of banning and allowing) The tense standoff between the U.S. government and the Silicon Valley AI giant has taken a dramatic turn. U.S. President Donald Trump revealed in an exclusive interview on CNBC's "Squawk Box" on Tuesday that an agreement allowing the Department of Defense (DOD) to use Anthropic's AI models "is possible." In the interview, Trump praised the startup he had personally ordered to be banned: "They came to the White House a few days ago, and we had a very pleasant meeting. I think they are shaping up. They are very smart, and I think they can be very useful." Once labeled a "national security threat," trouble arose from refusing to let AI become a lethal weapon Trump's friendly remarks stand in stark contrast to the public conflict between the Pentagon and Anthropic earlier this year. In March of this year, the U.S. Department of Defense officially labeled Anthropic a "supply chain risk," claiming that using the company's technology would threaten U.S. national security. This label mandated that all defense contractors provide certification to ensure that the Claude AI model was absolutely not used in their cooperation with the military. Subsequently, Trump posted on the social media platform Truth Social, demanding that all federal agencies "IMMEDIATELY CEASE" using Anthropic's technology, and stated firmly that his administration "will not do business with them anymore." It is understood that the trigger for the breakdown in negotiations occurred last September. At that time, Anthropic had just signed a $200 million contract with the Pentagon in July to deploy Claude on the DOD's GenAI.mil platform. However, the DOD demanded "unrestricted model access" for all lawful purposes; but Anthropic, which prioritizes AI safety, insisted that the military guarantee its technology would never be used for "fully autonomous weapon systems" or "domestic mass surveillance." The two sides remained deadlocked, eventually leading to the collapse of the partnership. "Breakthrough" after legal battles Faced with extreme pressure from the government, Anthropic did not back down. The company subsequently filed a lawsuit against the Trump administration in San Francisco and Washington D.C., attempting to overturn the blacklist ban. Currently, the executive order issued by Trump on social media has been temporarily blocked by a federal judge. Despite the tension in court, the relationship behind the scenes has begun to warm up significantly. Last Friday, Anthropic CEO Dario Amodei brought their powerful new AI model "Mythos" to the White House for a key meeting with high-level officials. The attendees of this closed-door meeting were of a very high level, including White House Chief of Staff Susie Wiles and U.S. Treasury Secretary Scott Bessent. A White House spokesperson later described the meeting as "productive and constructive." With Trump's public praise on Tuesday, whether Anthropic can win back the U.S. military's super-order while sticking to its moral bottom line will become the most watched focus in the tech world over the next few weeks.

資料狀態✓ 已擷取全文閱讀原文（動區 BlockTempo）

🔍歷史類似事件· 關鍵字 + 標的比對6 則

2026-04-24

Anthropic 承認 Claude「真的變笨」：三個工程配置失誤，已重置所有訂閱額度當補償

相似度 120%關鍵字 anthropic同分類 zh

2026-04-24

Claude 現可幫你叫 Uber、播 Spotify..連結八個日常生活 App，Anthropic 在下一盤大棋

相似度 120%關鍵字 anthropic同分類 zh

2026-04-23

OpenAI 執行長 Sam Altman 砲轟 Anthropic：恐懼行銷 Claude Mythos 只為壟斷 AI

相似度 120%關鍵字 anthropic同分類 zh

2026-04-22

彭博爆料 Claude Mythos 已遭未授權存取！Anthropic 最難防的破口永遠是「人」

相似度 120%關鍵字 anthropic同分類 zh

2026-04-22

在被投擲汽油彈後》Altman 稱末日論助長了 AI 產業的暴力，隨後捅了 Anthropic 一刀

相似度 120%關鍵字 anthropic同分類 zh

2026-04-22

汽油彈砸完》Altman 開口談末日論的代價，又拿 Anthropic 開了刀

相似度 120%關鍵字 anthropic同分類 zh

💡 目前用關鍵字 + 標的比對（MVP）· 之後會升級為 embedding 語意搜尋

原始資訊

ID：8c9dc07bdc

來源：動區 BlockTempo

發佈：2026-04-21 12:42:53

分類：zh_news · 導出分類 zh

標的：未指定

社群投票：+0 / −0 · ⭐ 0 重要 · 💬 0 留言