How many hours a week do employees at AI companies work?

According to testimonies from former employees, at OpenAI they typically work 50 to 60 hours a week, regularly exceed 70 hours, and reach up to 90 hours during sprints.

Why doesn't AI shorten working hours even though it increases productivity?

MIT researcher Neil Thompson explains that any time saved is quickly consumed by adapting to new processes and verifying AI outputs, so new work simply appears instead of a shorter workweek.

What does research say about AI's impact on working hours?

A Harvard Business School study showed that employees who use AI work more hours and broaden the scope of their jobs, while a Berkeley study found they work at a faster pace and take on more tasks.

Have AI companies implemented the four-day workweek they advocate?

No. Although OpenAI invited other companies to test a four-day workweek in early 2026, a former employee confirmed to the BBC that the company never applied that model to its own workers.

They Promise a Shorter Workweek, but Work 90 Hours: The Irony of the AI Industry

U.I. · Prije 17h | 3 min

While leaders at OpenAI, Meta, and Google herald the imminent arrival of a four-day workweek, their employees testify to exhausting sprints and a work culture that devours free time.

Summary

AI companies promise a four-day workweek, but their employees work up to 90 hours a week.
Research from Berkeley and Harvard shows that AI increases the scope of work instead of reducing it.
Employees who save 11 hours a week thanks to AI spend nearly 6.5 of those hours fixing errors and maintaining AI processes.
The CEO of Cerebras stated that it's impossible to achieve greatness by working 38 hours a week and having work-life balance.

Make 5MIN.hr a preferred source

Tech giants pouring billions into artificial intelligence development have been promising a radical reduction in working hours for years. Back in 2022, a Google engineering director predicted that AI would bring about a four-day workweek by 2025. In early 2026, OpenAI invited companies to test such a model without pay cuts. Yet the reality inside AI labs tells a very different story.

A former OpenAI employee told the BBC that they regularly worked at least 70 hours a week, and during intense "sprints" before product launches, the workweek often exceeded 90 hours. It's similar at Anthropic, whose Claude chatbot can work autonomously for seven hours without a break, but the humans developing it don't have that luxury. According to Gizmodo, OpenAI employees typically work between 50 and 60 hours a week, even outside of crunch periods.

A Work Culture from the Last Century

OpenAI president Greg Brockman has documented his own 90-hour workweeks, and Andrew Feldman, co-founder and CEO of AI chip company Cerebras, told Gizmodo: "This notion that somehow you can achieve greatness, you can build something extraordinary by working 38 hours a week and having work-life balance, that is mind-boggling to me."

Meta employees describe being moved to high-priority AI projects this year without any choice, where working late into the night and on weekends is routine. Ironically, Meta is simultaneously developing AI tools intended to take over some software engineering tasks, but these very projects are adding to the workload of teams. The company has also laid off about ten percent of its workforce, and Mark Zuckerberg has said that AI will enable a smaller number of employees to do more work.

Where Does the Saved Time Go?

Research shows that AI doesn't reduce work; it expands it. A study from the University of California, Berkeley, conducted over eight months, found that employees worked at a faster pace, took on more tasks, and extended their workdays. Harvard Business School published research earlier this year showing that employees who use AI simply work more hours and broaden the scope of their jobs, rather than cutting their working time.

Neil Thompson, a researcher at MIT, explains the paradox: "People think that 20 percent less work will mean a four-day workweek. In reality, new work just appears." A report from the AI Work Institute further confirms this pattern: some employees report saving up to 11 hours a week thanks to AI, but they spend about 6.5 of those hours maintaining AI workflows and fixing errors the tools generate.

A former Google employee said that due to the redirection of computing resources to AI projects, they increasingly worked late into the night. After leaving the company, their sleep and health improved significantly. While Anthropic CEO Dario Amodei warns that productivity gains could lead to more layoffs, the question hanging in the air is: will companies ever use the gains from AI to lighten employees' loads, or will they just pile on more work?