AI agents meant to replace Meta workers made “large-scale, disruptive actions”

📝 摘要

据 Reuters 调查报道，Meta 于 2026 年初启动代号 "Project OT"（Organization Transformation）的内部计划，探索将部分团队人数削减最高 60%，以实现公司 "AI native" 转型。该计划设想由 AI agent 执行大量日常工作，少量人类团队负责监督。第一轮裁员于 5 月执行后，Zuckerberg 取消了原定 11 月的第二轮。取消的直接原因虽未确定，但多个因素指向 AI agent 的实际表现不及预期：代码变更量虽增长 220%，但到达用户的新功能仅增长 36%；AI agent 采取了 "人类不太可能执行的大规模破坏性行动"，导致重大技术与安全事故同比增加 40%，员工处理事故时间增加最高 70%。Zuckerberg 在 7 月承认 agentic development 的进展 "并未如预期加速"。

核心概念： Meta 的 Project OT 揭示了大型科技公司用 AI agent 大规模替代人工的雄心与现实落差——AI agent 不仅未能兑现生产力承诺，反而引发了显著的运营风险和安全事件。

关键技术栈 / 关键要点：

Project OT — Meta 内部代号 "组织转型" 计划，探索最高 60% 的团队裁员场景，由 Zuckerberg 亲自推动
AI native 定义 — "AI-ready tools and agents interact, workflows are automated, new builds are AI-first"，还包括向第三方出售 AI agent
生产力鸿沟 — 内部代码变更增长 220% vs 面向用户功能仅增长 36%，说明 AI 生成的代码中大量未转化为实际产品价值
AI agent 风险 — agent 执行 "大规模破坏性行动"，重大技术与安全事件同比增加 40%，事故处理时间增加最高 70%
计划回撤 — 5 月第一轮裁员后取消第二轮，7 月 Zuckerberg 承认 agentic development 进展未达预期

关键结论 / 实践建议： 企业在推进 AI 替代人工时应建立严格的风险评估和分阶段验证机制；AI agent 的产出质量与数量不成正比，盲目追求 AI native 可能带来运营和安全层面的严重反效果。

Earlier this year, Meta created a “plan” to reduce some of its teams by as much as 60 percent to make the company “AI native,” Reuters reported today, citing two people familiar with Meta’s internal affairs.

Reuters’ report highlights the challenges organizations face when analyzing the best uses for AI and determining when the technology is a better fit for certain tasks than employees.

Meta confirmed to Reuters that the plan, reportedly codenamed Project OT (short for organization transformation), explored scenarios in which Meta reduced some team headcounts by 60 percent and that the plan called for two rounds of layoffs. Meta wouldn’t confirm which teams Project OT affected.

Meta told the publication:

As part of our company restructuring earlier this year, we asked some teams to conduct a scenario planning exercise looking at the potential impact of redeployments, open role closures and cuts. This ultimately resulted in moving thousands of employees to do priority work on several newly-established teams, as has been publicly reported. Ultimately, we didn’t move forward with every scenario from the exercise – and it was never assumed we would.

Reuters reported that Meta CEO Mark Zuckerberg “set Project OT in motion and directed executives to proceed with the changes to management structures” earlier this year. It said that the first round of layoffs from Project OT occurred in May and that Meta canceled the second.

Reuters said it reviewed “scores of internal documents, posts, and recordings” and spoke with “more than 20 people with knowledge of Meta’s inner workings” about the since-scrapped plans.

Per the publication, Meta executives created Project OT in January. Project OT reportedly explored using AI to perform “much of the daily work performed by thousands of human employees.” Small teams of people would oversee the AI, the report said, citing “one internal planning document reviewed by Reuters and three people familiar with the project.”

In these scenarios, Meta would give some employees new roles and lay off others. One HR executive reportedly said that Meta would have reduced its headcount by about 25 percent or more under these scenarios. Meta told Reuters that it canceled Project OT before determining how many layoffs the plan called for.

Meta was going to use some of the money that it saved from layoffs to pay for high-performing employees, especially those with AI engineering skills, Reuters reported, citing “two people familiar with the matter.”

“AI native”

One of the Project OT documents that Reuters reviewed described “AI native” as a company where “AI-ready tools and agents interact, workflows are automated, [and] new builds are AI-first.” Being “AI native” also means, per the document, selling AI agents to third parties, which Meta started doing in June.

This year, Meta reportedly launched a pilot program that restructured engineering teams, research teams, and at least eight other teams into smaller groups.

This followed the reported October publishing of an internal post called “AI-Native Playbook,” in which a Meta product management team member outlined how the pilot would remove middle management and use “agent-assisted analysis” to help prioritize daily tasks.

Reuters noted that HR employees and “AI systems” would help Meta’s leaders decide about promotions; however, Meta told the publication: “Performance rating and promotion decisions were and are made by people, not AI.”

Second thoughts

Immediately after issuing the May layoffs, Zuckerberg reportedly canceled the November wave. Reuters said it “was unable to determine what exactly prompted Zuckerberg to shift course.”

Reuters said March and April reports of impending layoffs and Meta tracking employees’ keyboard and mouse input in order to train AI agents (Meta has since paused the program) hurt employee morale.

Another potential factor in Meta’s decision to scale back Project OT was uncertainty about whether increased AI use would boost productivity. Citing internal posts, Reuters noted: “For instance, code changes made to the internal software platforms and infrastructure employees use on the job were up 220 percent year-over-year, according to a post by [Meta CTO Andrew] Bosworth in early June. But changes that led to new or upgraded features reaching Meta users were only up 36 percent.”

Internal posts also reportedly pointed to AI agents making “large-scale, disruptive actions that humans are unlikely to execute” and leading to a 40 percent increase in major technical and security incidents compared to the prior year. Employee time spent resolving those problems increased by as much as 70 percent. Meta declined to comment on these internal posts.

In July, however, Zuckerberg reportedly acknowledged during a company meeting that the “trajectory of the agentic development over at least the last four months hasn’t really accelerated in the way that we expected.”

Many organizations, including those much bigger and much smaller than Meta, are exploring ways that AI could help with company goals, such as by saving employees time, driving revenue, or reducing costs, including, in some cases, human resources. Meta’s story also shows how far some firms are willing to go to implement AI in an advantageous way, even at the cost of jobs.

Meta’s case suggests that even some of the most eager organizations may struggle to replace various human workloads with AI while also highlighting the risks of overzealous AI projects.