📄 中文摘要
2026年2月末，五角大楼要求Anthropic取消Claude在大规模监视和自主武器决策上的使用限制，Anthropic拒绝了这一要求。随即，特朗普下令在所有美国政府机构中逐步淘汰Anthropic的使用。就在几小时后，OpenAI签署了Anthropic放弃的合同。这一事件凸显了两家公司在技术应用伦理上的重大分歧，Anthropic明确禁止其技术用于监视或自主致命决策，而OpenAI则选择接受这一合同。
Anthropic 拒绝了 2 亿美元的五角大楼合同，OpenAI 在几小时后签署了合同
📄 English Summary
Anthropic Refused a $200M Pentagon Contract Over Two Use Cases. OpenAI Signed Hours Later.
In late February 2026, the Pentagon requested Anthropic to lift usage restrictions on Claude for mass surveillance and autonomous weapons decisions, which Anthropic declined. The following day, Trump ordered a six-month phase-out of Anthropic across all U.S. government agencies. Hours later, OpenAI signed the contract that Anthropic had walked away from. This incident highlights a significant ethical divide between the two companies, with Anthropic explicitly prohibiting its technology from being used for surveillance or autonomous lethal decisions, while OpenAI opted to accept the contract.