📄 中文摘要
2026年2月15日，关于“安全与伦理AI”的理想与全球防务的现实发生了冲突。OpenAI、Google和xAI悄然修改服务条款，以允许其模型在军队中使用，而以“AI安全”为原则的Anthropic则与五角大楼发生了直接冲突。争议的核心在于，五角大楼希望顶尖模型能够在毫秒内评估目标并提出消除威胁的方案，而这涉及到Claude模型在自主武器决策过程中的直接参与。
伦理前沿2026：Anthropic与五角大楼的对抗及人工智能中立性的终结
出处: Etická fronta 2026: Anthropic vs. Pentagon a konec AI neutrality
发布: 2026年2月15日
📄 English Summary
Etická fronta 2026: Anthropic vs. Pentagon a konec AI neutrality
February 15, 2026, marks a pivotal moment where utopian visions of 'safe and ethical AI' collided with the harsh realities of global defense. While OpenAI, Google, and xAI quietly adjusted their terms of service to permit military deployment of their models, Anthropic, a company founded on 'AI Safety' principles, found itself in direct conflict with the Pentagon. The crux of the dispute lies in the Pentagon's push for advanced models to evaluate targets and propose threat eliminations within milliseconds, directly involving the Claude model in decision-making processes for autonomous weapons.