在周五晚上，OpenAI在其时任CEO Sam Altman宣布接替 Anthropic vacated（放弃）在美国国防部的AI供应角色后，数小时内舆论即猛烈反弹。Anthropic CEO Dario Amodei 坚持 Anthropic 的模型不得用于对美国人进行大规模监控或开发完全自主武器，这一立场激怒了美国国防部长 Pete Hegseth。与此同时，美国与以色列对伊朗发动首次齐射攻击，且当地当局随后称一所女子学校被毁、死亡人数超过160人。批评者很快将此事解读为 OpenAI 为了政府业务而牺牲普通民众与海外平民的隐私与伦理考量。
面对这一局势，Sam Altman 花了数日进行辩解，并在周一通过 X 发布声明，承诺修订协议以禁止对美国公民实施国内监控，并称此前的国防部合作看起来“机会主义且草率”。在这段期间，Anthropic 的 Claude 在应用商店下载量超越了 ChatGPT；周一早晨的下载峰值甚至导致服务短暂崩溃，影响数千用户。大量用户公开取消付费 ChatGPT 订阅并切换至 Claude，Claude 自此前在非技术群体中的认知和使用率明显落后于 ChatGPT，转而在当日早晨重新登顶苹果设备新下载榜首。
这篇报道将这一现象解读为消费者对 Trump 政策（包括关税与移民治理）不满下，对AI公司施加政治压力的市场化表达。Anthropic 的立场也赢得了部分竞争对手员工支持；Hegseth 先前将其列为供应链风险后，其他国防承包商据说已停止使用其工具。但 Anthropic 的商业现实依然矛盾：它仍为 Palantir Technologies 的数据与监控工具提供底层AI，曾向国防部提交过自动无人机集群方案，并在周三重启与国防部谈判。对于依赖消费者付费的公司而言，Claude 的激增向其他高管发出信号：如果敢划定道德红线，消费者可能会以可见的市场收益予以回应。
Within hours of OpenAI CEO Sam Altman announcing that OpenAI would step into the U.S. Department of Defense role vacated earlier that Friday evening by Anthropic, public backlash intensified rapidly. Anthropic CEO Dario Amodei had refused to allow its models to be used for mass surveillance of Americans or fully autonomous weapons, which infuriated Secretary of Defense Pete Hegseth; that same night, U.S. and Israeli forces opened an Iran strike with its first salvo, and local authorities later said a girls’ school was destroyed and over 160 people were killed. The reaction framed OpenAI’s move as a tradeoff of ordinary and foreign civilians’ rights for government contracts.
Sam Altman spent several days defending the decision and on Monday posted on X that the company would revise its agreement to prevent domestic surveillance, while admitting the original Pentagon arrangement appeared opportunistic and sloppy. In the same interval, Anthropic’s Claude overtook ChatGPT in app downloads; on Monday morning the spike was large enough to temporarily crash services for thousands of users. Many people canceled paid ChatGPT subscriptions and switched to Claude, and Claude, which had previously lagged well behind ChatGPT among non-tech users, was still the top-ranking new download on Apple devices that morning.
Bloomberg describes this as one of the first large-market demonstrations of consumers politically punishing AI firms amid frustration with Trump-era policies, including tariffs and immigration enforcement. Anthropic’s stance also triggered worker support within rival firms, and after Hegseth labeled it a supply-chain concern, other defense contractors were reported to have dropped its tools. Yet Anthropic remains commercially entangled: it still supplies core AI for Palantir Technologies surveillance systems, had earlier proposed autonomous drone-swarm technology to the Pentagon, and by Wednesday had resumed Pentagon talks. For firms that depend mainly on consumer choice, Claude’s surge suggests that setting a public ethics boundary can still create measurable commercial upside.