Meta's Morale Is Near a 20-Year Low, and the AI Pivot Is Why

TL;DR

Meta CTO Andrew Bosworth told staff that morale is near its worst in 20 years, weeks after the company cut about 8,000 jobs (roughly 10 percent of its 78,000-person workforce) and posted $26.8 billion in quarterly profit. The trigger is the AI pivot: thousands of engineers were swept into a new Applied AI unit with little say, and since April Meta has been logging employees' keystrokes and screenshots to train AI agents. The morale fix so far is bigger budgets for travel, events, and office snacks. Employees were not impressed.

Record profit, record gloom

The timing is the story. In May 2026 Meta laid off about 8,000 people, roughly 10 percent of a workforce of around 78,000, and cancelled another 6,000 open roles. That came just weeks after one of its strongest quarters ever: $26.8 billion in net income on $56.3 billion of revenue, up about 33 percent year over year.

Asked about the mood internally, Bosworth did not really deny it. His read, per reporting on the internal Q&A, was that morale is "maybe not the worst it's ever been in 20 years here, but it's probably up there." For a company that already ran an 11,000-person cut in 2022 and a 10,000-person "year of efficiency" in 2023, "up there" is a grim bar to be near.

The AI reorg nobody asked for

The deeper driver is not the headcount math, it is how the AI pivot landed. In March 2026 Meta stood up a new Applied AI unit and swept roughly 6,500 engineers and product managers into it, many with little input and little clarity about what it meant for their careers. One worker quoted in the reporting called the division "a gulag."

That is the part builders should sit with. Being told overnight that your team is now an AI team, with no say in the move and no roadmap, is a fast way to convert senior talent into flight risks. Reorgs are normal. Reorgs framed as "everyone is AI now, figure it out" are how you lose the people who had options.

Training your replacement, literally

Then there is the surveillance. Since April, Meta has run software on US employees' laptops that logs keystrokes, clicks, and periodic screenshots, explicitly to generate training data for AI agents. More than 1,500 workers signed a petition against it. Meta later softened the policy to let employees pause the tracking in 30-minute windows.

Read that back slowly: a company doing AI layoffs is recording how its remaining staff work in order to train agents that do that work. Whatever the intent, the optics are employees instrumenting their own automation. It is also a preview of a pattern other companies will copy, and a consent and data-provenance minefield that has barely been tested.

The snack-budget morale fix

Meta's visible response has been perks. The company increased budgets for travel, events, and office snacks as a gesture toward improving day-to-day conditions. The reaction, internally and online, was mockery. Free almonds do not address being reassigned against your will, surveilled, or watching colleagues get cut in a record-profit year.

This is the oldest mistake in people management dressed in AI-era clothes: treating a trust-and-purpose problem as an amenities problem. Perks are a multiplier on good morale, not a substitute for it.

Why a builder should care

The Meta morale story is an AI story. This is what "AI transformation" looks like from the inside of a top lab: forced reassignment into AI work, monitoring to feed agent training, and layoffs running concurrently with record earnings. If you are weighing a big-lab role, the Applied AI dynamic and the train-your-replacement pattern are the parts the recruiter will not lead with.

For anyone building agents, the data angle is the one to watch. Harvesting employee interaction telemetry to train automation is going to get copied, and the consent, provenance, and morale costs Meta is paying now are the bill the rest of the industry has not yet seen.

Key Takeaways

• Profit did not buy goodwill. Roughly 8,000 layoffs landed weeks after a $26.8 billion quarter, and Meta's own CTO put morale near a 20-year low.
• The AI reorg is the real wound. About 6,500 staff were moved into Applied AI with little say, with one calling it "a gulag." Forced AI reassignment is a talent-retention disaster.
• Employees are training the agents. Keystroke and screenshot logging on US laptops to build AI agents drew a 1,500-signature petition, and a partial walk-back to 30-minute pause windows.
• Snacks are not a strategy. Bigger perk budgets read as superficial against involuntary reassignment and surveillance, and got mocked accordingly.
• This pattern will spread. Surveil-to-train plus AI-driven cuts is a template other companies will copy, with consent and provenance questions still unresolved.

Sources: IBTimes UK, Yahoo Finance, Fortune via Yahoo, Mashable