Andrew Bosworth admits Meta morale is at rock bottom after restructuring fallout

Meta’s CTO Andrew Bosworth went from “you should quit if you feel that way” to acknowledging that the damage is real. In a memo and a meeting with employees this month, Bosworth said morale is “probably one of the worst it’s ever been” and that Meta had done “an atrocious job” with its recent restructuring. He also said the company had “undermined the trust” employees need to believe their expertise and contribution will be valued.

That reversal matters because Meta spent the past few years trying to become a lean, fast, high-pressure machine. Early last year, Bosworth’s message to staff was blunt: if people felt treated poorly or questioned controversial changes, they “can leave,” or they should “disagree and commit.” The company was reinforcing a specific culture, the kind that tolerates little internal debate and expects employees to sprint through discomfort. But now the sprint has hit a wall. The source describes Meta’s workforce as being at a breaking point, with unrest in the UK, surveillance fears tied to AI model training, and employee attempts to force attention to their frustrations.

The employee pressure is not subtle. Employees in the UK are trying to form a labor union and are decrying executives’ “cruel and shortsighted behaviors.” More than 1,600 workers have signed a petition demanding Meta stop tracking employees’ keystrokes to improve its AI models. And the internal mood is loud enough to leak into the public internet: Wired reported this month that a frustrated employee hijacked a livestreamed meeting with a profanity-laced outburst aimed at an executive. The same report says another worker compared working in a new AI-training unit to the gulag. Others are reportedly so dejected they are praying to be laid off so they can at least leave with severance.

Bosworth’s comments arrive alongside a broader scrambling for damage control from other top leaders. Chief Product Officer Chris Cox acknowledged the “insanity of this company” and called the environment “difficult” and “brutal.” CEO Mark Zuckerberg admitted “we’ve made mistakes.” The source stresses a credibility gap: it’s “unthinkable” that Zuckerberg could be a credible spokesman for change. That’s not because the memo language is new, it’s because the actions that created the current trust problem were also senior-leadership actions.

The digging started with layoffs. Meta laid off 11,000 people in late 2022, a move Zuckerberg was at least apologetic about. The company then slashed another 10,000 jobs the next spring during what Zuckerberg called a “year of efficiency.” In 2025, Zuckerberg said another 3,600 job cuts were to get rid of “low performers,” which the source notes effectively torpedoed some workers’ job searches even though many had received good performance reviews. Then came the psychological purgatory. In March, news leaked that more cuts were coming, but Meta did not confirm the cuts for weeks. It also did not notify employees affected until May, leaving people stuck in uncertainty for about two months.

In April, while employees were in that limbo, Meta announced it would start tracking employees’ keystrokes, which the source frames as stoking fears that the company wanted to automate workers’ output. In May, Meta laid off 8,000 employees and reassigned another 7,000 to menial jobs involving training AI. Meta declined to comment on the story. Cox’s comments to Instagram employees this month are also included to highlight the gap between internal messaging and employee reality, including his marathon-in-a-hailstorm analogy and a profanity-laced description of the situation. If you’re trying to run a high-performing product culture, these are not optics problems. They are engagement and retention problems.

So why the mea culpa now? The source offers a few possibilities grounded in incentives and outcomes. One is that anger, dissatisfaction, and open rebellion were harming productivity. Another is that Meta’s dysfunction had become too public, turning into a reputation liability with investors. Or there may be a simpler truth underneath: perhaps leadership finally acknowledged that the “hard-charging management style” wasn’t delivering what it was built to deliver. The stated goal was to innovate faster and catch up in AI to competitors like OpenAI, Anthropic, and Google. Yet the timeline described suggests Meta’s AI delivery has lagged. The company delayed and ultimately never released what was supposed to be its flagship AI model after engineers reportedly struggled to improve capabilities. It also repeatedly pushed back another model’s rollout to developers.

That’s where management research enters the story, and it’s the bridge from employee pain to strategic risk. Decades of management research suggest fear and instability hemorrhage star employees, make recruiting harder, and suppress the creative risk-taking needed for breakthroughs. Sandra Sucher, a professor of management practice at Harvard Business School, is quoted in the source calling it “a classic example of chickens coming home to roost,” saying Meta “almost systematically destroyed trust” and is now trying to dig itself out. She also suggests executives are making the right move by acknowledging what they did wrong, but warns that real trust repair requires more than partial concessions.

Meta’s current steps include promises to reduce the size of teams managers oversee, scale back the keystroke-monitoring program, and increase budgets for social events. Employees reassigned to AI training roles now have the opportunity to opt into a different role of their choice, the company announced internally this week. But Sucher’s core point is that the first step is a real apology that includes the word “sorry,” and most importantly a credible commitment that leadership won’t keep making the same fundamental error: treating employees as if they’re not human beings who deserve respect and care. The source underlines the practical logic of that: workers do not simply watch layoffs, surveillance, and unwanted transfers and then magically return to peak performance.

For the broader tech industry, this matters beyond Meta’s walls. The source ties Meta’s aggressive management era to a turning point in Silicon Valley, pointing back to November 2022 when Meta became the first tech giant to conduct mass layoffs. The question now is whether Bosworth’s shift is a real course correction that changes how leadership governs, or a temporary pause before returning to harsher methods. Similar firms are already making different choices, with Google and Microsoft opting for voluntary buyouts over mass layoffs in recent months, and Zuckerberg promising a period of stability, holding off on any more big job cuts at least through the end of the year. If Meta’s new approach sticks, it could redefine what “efficiency” looks like in AI competition. If it doesn’t, it becomes a case study in what happens when speed is built on trust debt.