自从生成式AI热潮起，科技领导者一边推广一边放大恐惧：Sam Altman、Dario Amodei 等先前称ChatGPT相关技术可能带来生存风险，而后又将焦点转向 AI 取代工作与加剧资安威胁。Anthropic 的 Mythos 被描述为可在多种作业系统与浏览器中发现零时差漏洞（zero-day），公司仅开放给 Apple、Amazon 等少数预先审核企业，使「太危险而不宜外流」的叙事更具说服力。

主要论证带有强烈数字化冲击：Amodei 在2026年1月称未来1至5年间可能消灭一半入门白领职位；Mustafa Suleyman 在2月又称全部专业任务有望在18个月内被自动化。OpenAI 于4月9日（星期一）发布13页的《Industry Policy for the Intelligence Age》，提出公共财富基金、机器人税与四天工作制等构想。文中也提到截至目前，南韩是唯一已尝试自动化课税的国家，显示政策实效仍未可知。

但整体就业与产出数据并未支持全面崩溃论。Oxford Economics 研究简报指出，AI 大规模失业恐慌依赖一连串假设；AI更可能辅助而非全面取代劳动。NBER于3月调查6,000名美欧企业高层，发现过去3年有90%企业未感受到AI对就业的影响；同时在资讯产业，聘用率与裁员率都在上升。文章最后指出，Anthropic 筹备IPO与OpenAI面临管理、资金消耗与监管压力，为何其对外「预警」与「解方」容易被解读为同时的市场沟通策略。

Since the generative AI boom began, technology leaders have simultaneously marketed systems and amplified fear: Sam Altman, Dario Amodei, and others earlier warned ChatGPT-related technology could pose existential risk, and now the narrative has shifted to AI as a job killer and a cybersecurity threat. Anthropic says its upcoming Mythos model can detect zero-day vulnerabilities across multiple operating systems and browsers, and limits early access to a handful of vetted firms such as Apple and Amazon, creating a narrative of controlled distribution because the model is too dangerous to release broadly.

The arguments are quantitatively intense: in January, Amodei claimed that in the next 1 to 5 years AI could eliminate half of entry-level white-collar jobs; in February, former Microsoft AI executive Mustafa Suleyman warned that all professional tasks could be automated in only 18 months. OpenAI’s 13-page document, Industry Policy for the Intelligence Age, published on Monday, April 9, 2026, proposes measures such as a public wealth fund, robot taxes, and a four-day workweek. The article notes that South Korea is the only country so far to have attempted levies on automation, highlighting policy uncertainty.

Yet the evidence suggests a weaker case for collapse: an Oxford Economics briefing says the AI doomsday-job-loss chain depends on many assumptions and that AI is often augmenting rather than replacing workers. A March NBER survey of 6,000 executives in the US and Europe reports that 90% of firms saw no impact from AI on employment over the past 3 years, while hiring in the information sector rose along with layoffs. The piece argues this disconnect helps explain why firms can still weaponize AI narratives—such as Block’s 4,000 layoff episode—while leaders call for action, especially as Anthropic prepares for an IPO and OpenAI faces management and cash-burn scrutiny.