Meta has scrapped the second wave of layoffs planned for November 2026 after internal criticism. That’s according to reporting by the news agency Reuters, based on internal company documents. According to those documents, at least 20 percent of the roughly 79,000 positions were originally slated for elimination – in the end, only the first wave of ten percent, effective in May, went ahead.
Internal numbers show a failing agent strategy
The reversal traces back to “Project OT,” a program launched in January 2026 that CEO Mark Zuckerberg and his leadership team devised during a retreat in Hawaii. The goal was an “AI-native” company in which autonomous software agents would take over much of the work currently done by human employees. In some planning scenarios, specific teams were to shrink by as much as 60 percent – far more than the ten percent Meta actually carried out in May.
Internal metrics painted a different picture than hoped: code changes rose 220 percent year over year, but features that actually reached users increased only 36 percent. At the same time, major technical incidents rose 40 percent, and the effort needed to fix them climbed 70 percent. At an internal meeting in July, Zuckerberg acknowledged that agent technology had not progressed as expected over the prior four months.
Employees push back against surveillance and job fears
Resistance to the plans grew after Meta installed software on US employees’ computers that logs keystrokes and mouse movements to generate training data for its in-house AI agents. Many employees responded with angry posts and gallows humor on the internal network Workplace – some illustrated their fear of training their own replacements with images of elephants, a nod to the elephant in the room. Meta’s twice-yearly sentiment survey fell from 74 to 55 percent favorable over this period.
On the night of May 19, just hours before the first wave of layoffs began, Zuckerberg decided, according to the internal documents, not to proceed with the second round planned for November. The first wave, affecting roughly 8,000 positions in May, had already prompted a lawsuit in July from 26 employees who accused Meta of using an algorithm to select protected employees for layoffs.
Meta confirms the program, disputes the details
A company spokesperson confirmed to Reuters that “Project OT” existed but denied that every scenario examined had been carried out. Instead, thousands of employees were shifted to newly created, prioritized teams – an outcome the company says was never assumed to be certain. On the allegations from July’s lawsuit, Meta reiterated that performance ratings and promotion decisions are made by people, not software. Zuckerberg told remaining staff he does not expect any further company-wide layoffs for the rest of 2026.
The reversal changes nothing about Meta’s massive AI infrastructure spending: the company recently guided to a budget of up to $145 billion for 2026, well above the roughly $72 billion it spent the prior year.
Whether the admitted failure of the internal agent strategy will also affect the pending lawsuit from the 26 plaintiffs, who accuse Meta of delegating the May layoff selection to software, remains to be seen. For other tech companies also shrinking their workforce by pointing to AI efficiency, the case offers a first, publicly documented example of how quickly such plans can collide with reality once the promised productivity fails to show up.
