'AI Will Free Us' — But Workers Inside AI Companies Say They Log Up to 90-Hour Weeks

Inside the 90-hour week at OpenAI and Anthropic

Employees at leading AI labs describe grueling schedules that far exceed the traditional 40-hour workweek. A former OpenAI technical employee told the BBC they worked a minimum of 70 hours per week, and during "sprints" — the period before a major feature launch — hours climbed to 90 per week .

This pattern is widespread across the industry. An ex-OpenAI employee now at another AI startup described the culture as marked by frequent "crisis meetings," mandatory weekend work, and "super cut-throat" performance reviews where colleagues are suddenly let go . The BBC also documented AI startups openly advertising 70-hour weeks as a selling point, with one job ad warning candidates not to apply unless they are "excited about working ~70 hrs/week in person" .

Bloomberg reported on June 26, 2026 that across Silicon Valley, the AI productivity boom is producing more anxiety and longer hours, not less work . At Anthropic, engineering lead Fiona Fung said on Lenny's Podcast that agentic AI use had become so pervasive it was making employees' work "a lonely experience" .

The UC Berkeley study: AI didn't free up time, it expanded the job

An eight-month ethnographic study by UC Berkeley Haas researchers Aruna Ranganathan and Xingqi Maggie Ye, published in the Harvard Business Review (February 2026), found that generative AI did not reduce anyone's workload . At a 200-person tech company where AI adoption was voluntary, researchers observed that employees using AI tools:

• Worked at a faster pace
• Took on a broader scope of tasks — product managers began writing code, designers took on engineering work
• Extended work into more hours of the day, often without being asked

The researchers concluded that AI "didn't free up time — it expanded the job" . The efficiency gains were captured by management and translated into expanded responsibilities without additional headcount . An ActivTrak report cited by Fortune found that time spent on job responsibilities increased by 27% to 346% after AI adoption .

AI-driven layoffs are rising sharply — and companies are explicit about the reason

According to the July 2026 Challenger, Gray & Christmas report, AI was cited as the leading reason for job cuts for the fifth consecutive month . Tech sector layoffs in the first half of 2026 were up roughly 83% year-over-year, with tech companies accounting for more than 30% of all U.S. job cuts in 2026 .

Major firms have explicitly cited AI investment or AI-driven productivity as a reason for workforce reductions:

• Meta cut roughly 8,000 jobs (10% of its workforce) while spending $135 billion on AI in 2026 alone
• Block (Square, CashApp) laid off nearly half its staff — CEO Jack Dorsey told shareholders AI "fundamentally changes what it means to build and run a company"
• Amazon, Microsoft, and Oracle have all cited AI as a factor in job cuts

By May 2026, 40% of all tech layoff announcements named AI as a factor, up from just 7% in January . So far in 2026, AI has been cited in more than 112,000 job cut announcements .

The AI training gap

Despite the rapid deployment of AI tools, only about one in three workers have received employer-provided AI training, and most of that training focuses on basic skills rather than deeper integration or upskilling . This leaves the majority of workers expected to use AI tools without adequate support, even as their workloads intensify.

The broader picture: a productivity paradox

The accumulating evidence points to a fundamental contradiction at the heart of the AI boom. The technology that was supposed to liberate workers is, inside the companies building it, producing:

• Longer hours (70-to-90-hour weeks are normalized)
• Expanded responsibilities (AI tools blur job boundaries and increase multitasking)
• Job insecurity (AI is now the No. 1 reason cited for layoffs)
• Inadequate training (most workers are left to figure it out alone)

The BBC investigation, along with supporting reporting from Bloomberg, Fortune, and the Harvard Business Review, shows that the AI boom has produced a work intensification paradox inside the very companies promoting AI as a labor-saving technology .