America Sets No Limit on Your Working Week While Britain Caps It at 48 Hours

There are no limits in the United States on how many hours a person over the age of 16 can work, while in the United Kingdom and Europe laws cap a working week at 48 hours including overtime, according to BBC News. That difference has stopped being an abstraction. Tech workers told the BBC that sprints at leading AI companies can stretch for weeks and top 90 hours of work in a seven day period. At Metaintro, we track roughly 50 million job postings, and the useful question is not which system is better but what each one actually gives you.

What Does the Law Actually Say in Each Place?

In the US, federal law regulates how you are paid for hours rather than how many you work. There is no statutory ceiling on weekly hours for an adult worker. The protection that exists is financial, in the form of overtime pay for eligible employees, which raises the cost of long hours without prohibiting them. Whether you qualify depends on your classification, and our guides to what full time exempt actually means and when exempt workers are legally doing hourly tasks cover where that line falls.

In the UK and across the EU, the approach is a hard cap. The working week is limited to 48 hours including overtime, which is a limit on the hours themselves rather than on their cost. The practical effect is that the ceiling exists whether or not your employer is willing to pay more.

For a salaried professional in the US, the consequence is stark. If you are exempt from overtime, there is no legal ceiling and no financial brake, which means the only real constraints are your employer's judgement and your own willingness to leave. Recent changes have narrowed that protection further, as our explainer on the 2026 overtime rule and what it means for salaried workers sets out, alongside new Labor Department guidance on overtime and bonus pay.

Why Is This Suddenly Relevant to AI Work?

Because the industry that promised shorter hours is producing some of the longest. Executives at companies investing heavily in AI have argued for years that the technology would reduce time spent working. An engineering director at Google predicted four years ago that AI would deliver a four day work week by 2025, and OpenAI formally urged companies to start testing a four day week with no change in pay, on the basis that AI would soon accelerate human labour enough to justify it.

The reported reality inside those companies runs the other way. A former OpenAI technical employee told the BBC the firm never actually trialled the four day week it suggested others should try while they were there, describing instead a culture of frequent crisis meetings, weekend work and "super cut-throat" performance reviews. The same person said they put in at least 70 hours a week, considerably more than in previous tech jobs.

The pattern is not confined to one company. At Meta, workers described being abruptly moved onto urgent AI teams in a practice they called being drafted, with one former employee saying "they just move you over" and describing a choice between accepting the transfer or resigning. Staff on those teams reported working into the night and at weekends and feeling on call when not working. Meta has since begun to walk back that approach, which we covered in the reversal on AI training role reassignments, though by then many had already been moved.

Is There Evidence That AI Expands Workloads?

Yes, and it is more specific than anecdote. A study out of the University of California, Berkeley followed hundreds of workers at a US technology company for eight months as they adopted AI tools, and found that employees "worked at a faster pace, took on a broader scope of tasks, and extended work into more hours of the day". Part of that expansion came from the need to constantly check the output of the tools.

Neil Thompson, an innovation scholar at the Massachusetts Institute of Technology, described the mechanism plainly, saying it "leads to a situation where even if there were real time savings, it would be sucked up by the changes, and implementing them, and making sure they worked". He added that "people assume that 20% less work means four-day weeks" but that "new work emerges".

That second point is the one worth internalising. Time saved on a task does not automatically become time off, because the space created tends to be filled either by the overhead of managing the tool or by additional work that people take on to demonstrate their value. Without a policy or a legal limit forcing the saving to be realised as time, it simply disappears into the week. Our coverage of what a shorter week could look like if AI gains were shared explores the alternative.

What Protections Do You Actually Have?

Fewer than most people assume, and the honest answer depends on where you work and how you are classified. In the US, an exempt salaried employee has no hours ceiling and no overtime entitlement, so the binding constraints are contractual and cultural rather than legal. In the UK and EU, the 48 hour cap exists as a genuine legal limit, though the picture varies by country and by whether an individual opt out applies.

The comparison is genuinely narrowing in some respects, as our analysis of UK work hours catching up with the US describes, and the UK has been reworking its framework more broadly, covered in our guide to the workers rights overhaul. Legal protection is not static, and assuming today's rules will hold is unwise in either jurisdiction.

The most reliable protections are the ones written into your own arrangement rather than the ones you inherit from statute. Clarity about on call expectations, about whether weekend availability is assumed, and about how sprint periods are defined and bounded will matter more day to day than the national framework. Where a legal cap exists, know whether you have signed anything that opts you out of it.

What Does This Mean for Your Career Decisions?

It means hours should be treated as a term of employment rather than a cultural detail you discover later. Two offers with identical pay are not equivalent if one carries a conventional week and the other routinely runs past 70 hours, since the second is a substantially lower hourly rate for work that is also harder to sustain. Converting the offer into an hourly figure before you accept is a simple discipline that changes decisions.

Ask concrete questions rather than atmospheric ones. How long the last sprint ran and how many hours it involved, whether people work weekends during a release, and what on call looks like in practice are all answerable in specifics. Vague reassurance about work life balance is not evidence, and the gap between a company's stated culture and its actual calendar is where most unpleasant surprises live.

The health consequences are not hypothetical either. One former Google employee told the BBC that his sleep and overall health improved after leaving, and said that heavy reliance on AI tools in large tech companies "creates a bad culture with excess pressure on engineers". Our coverage of how irregular hours affect the brain in shift work covers the physiological side, and our guides on working long hours and staff working long hours look at how to handle it in practice.

Geography is now a genuine variable in this decision. For roles that can be performed from more than one country, the applicable working time framework is part of the compensation package in a real sense, and for a remote worker the jurisdiction governing the contract may matter as much as the salary attached to it.

Why Does a Legal Cap Change Behaviour More Than Overtime Pay Does?

Because the two rules solve different problems. Overtime pay makes long hours expensive, which works well where the employer is choosing between paying one person more or hiring a second. It works far less well for salaried professionals who are exempt from it, since there the cost of an extra hour is effectively zero to the employer and entirely borne by the worker.

A hard cap removes that asymmetry by taking the decision out of the negotiation. Where hours are capped, an employer who needs more output has to hire, redistribute or descope, and none of those options quietly transfer the cost onto an individual. That is why the 48 hour framework produces different behaviour rather than merely different paperwork.

The trade off is real and worth acknowledging. Caps reduce flexibility for people who genuinely want to work intensively for a period, whether for equity upside, a product launch or their own ambition, and opt out provisions exist partly to accommodate that. The honest summary is that the US system maximises freedom to contract and the European system maximises protection, and which one serves you depends on whether you have the bargaining power to set your own limits.

How Do Sprints Turn Into Permanent Overwork?

Through a definitional drift that is easy to miss while it happens. A sprint is meant to be a bounded period of intense effort before a release, and on that definition it is a normal feature of building software. The problem arises when the bounding disappears. At leading AI companies, sprints can stretch on for many weeks and top 90 hours in a seven day period, which is no longer a sprint in any meaningful sense.

The tell is whether there is a defined end and a recovery period after it. A sprint with a date attached and a lighter fortnight afterwards is a workload management tool. A sprint that is followed immediately by the next one is a permanent operating model wearing a temporary label, and the language matters because it discourages people from objecting to something described as exceptional.

This is where AI adoption compounds the problem rather than easing it. If the freed capacity is absorbed by checking model outputs and by the additional scope that arrives with it, as the Berkeley research found, then the tools do not shorten the sprint, they raise the throughput expected during it. Our analysis of the widening gap between AI productivity and worker pay covers where that additional output ends up, and the coming burnout from supervising AI agents covers what it costs the people doing it.

Is the Long Hours Culture Specific to AI Companies?

Not entirely, though the intensity appears elevated. Investment banking, law and medicine have long involved extended hours, and the BBC noted that a 14 hour workday was not always the norm in technology, which for years looked more like a conventional office job. What has changed is the direction of travel inside an industry that publicly promised the opposite.

The distinction that matters is between sprints and steady state. The former OpenAI employee who described at least 70 hour weeks now works at an AI startup at closer to 50 to 60 hours outside sprints, which shows the range within the same broad sector. When you are evaluating an employer, the question is not whether intense periods happen but how long they last and how often they recur.

Related Articles

• AI and the Four Day Workweek
• The Case for AI Enabling a Shorter Working Week
• Could a Shorter Week Save Jobs From AI
• How Meta Staff Described the AI Reorganisation
• Why Workers Are Hoarding Paid Time Off
• The Real Reason Workers Are Maxing Out Their Leave
• What Leave Maxxing Is and Why It Is Spreading
• Companies Paying Workers Vacation Stipends
• Meta Careers Guide, Roles, Pay and Hiring Tips
• The US Labour Market Is Splitting Between AI Office Jobs and Hands On Hiring
• Tech Layoffs Tracker
• Adaptability Beats Raw Experience for Career Growth

People Also Asked

Q: Is there a legal maximum number of hours you can work in the US?

A: No. There is no limit in the United States on how many hours a person over the age of 16 can work. Federal law governs overtime pay for eligible employees rather than capping hours, and workers classified as exempt have neither a ceiling nor an overtime entitlement.

Q: How many hours can you legally work in the UK and EU?

A: The working week is capped at 48 hours including overtime under UK and European law. Details vary by country and individual opt outs exist in some cases, so it is worth checking whether you have signed one.

Q: Is AI actually reducing working hours?

A: The available evidence says no. A University of California, Berkeley study following hundreds of workers for eight months found they "worked at a faster pace, took on a broader scope of tasks, and extended work into more hours of the day". Tech workers described sprints topping 90 hours in a seven day period at leading AI firms.

Know your rights before you sign anything. Metaintro tracks roughly 50 million job postings across markets with very different working time rules, so you can compare what a role really asks of you, and you can create a free account to get matched with employers whose expectations match the week you actually want.