AI Lab Workers Debate Vacations as the Race for AGI Intensifies
A former OpenAI researcher's post sparked a public debate about vacation culture at AI labs like OpenAI and Anthropic, where high-intensity work norms are common.
A post by Miles Brundage, a former OpenAI policy research director, arguing that rest matters has ignited a public conversation about whether the intensifying AI race is eroding vacation culture at top labs. Employees from OpenAI, Anthropic, Google DeepMind, and Meta chimed in on X in late August 2026, some joking about how little time off they take and others defending their right to disconnect. The debate exposes a real tension at the heart of AI work: competitive urgency versus sustainable output.
What happened
| Data point | Detail |
|---|---|
| Post date | August 23, 2026 (X/Twitter) |
| Author | Miles Brundage, former OpenAI policy research director |
| PTO access (US info workers) | 95%, per US Bureau of Labor Statistics |
| Avg. tech company PTO offered | 19.8 days, per 2025 Alliant survey |
| 996 schedule | 9am to 9pm, six days a week |
Brundage’s post was straightforward: rest is important, and some AI labs may value it more than others. He did not name names, but the comment section filled up fast. An Anthropic employee shared a trip to the Edinburgh Fringe Festival. An OpenAI employee said the thread made them think they should book time off. OpenAI’s Jason Liu asked Anthropic’s Thariq Shihipar whether he agreed with Brundage. Shihipar’s reply: “Aren’t you always on vacation?”
Leo Gao, an OpenAI staffer, said he has “a tendency to let work consume all of his life” and finds that fits the culture there. Neither OpenAI nor Anthropic responded to interview requests.
The conversation spread beyond those two labs. Alex Imas, who leads AGI economics at Google DeepMind, wrote that he had just moved from academia and was “still figuring out what this ‘vacation’ thing is.” Angel investor Kevin Weimer pointed to a lab known for pushing limits: “Let’s talk about SpaceXAI. I don’t think they sleep.”
Why it matters
On paper, AI workers are not being denied time off. With 95% of US information workers having PTO access and an average of 19.8 days offered at tech firms, the formal policy is generous. The real issue is social pressure and the fear of falling behind in a field that moves fast enough that a two-week break can feel like missing a year’s worth of announcements.
One AI professional described keeping up with the field as “drinking from a fire hose that keeps getting bigger.” That framing captures the psychological cost. It is not that people cannot take leave; it is that the constant stream of model releases, benchmark updates, and product launches makes disconnecting feel costly.
Some workers have responded by adopting extreme schedules. The “996” pattern (9am to 9pm, six days a week) has emerged among some software engineers, along with dropping habits like alcohol to sharpen focus. Whether those choices reflect genuine passion or quiet coercion is the harder question this debate has not resolved.
Sam Altman’s position, stated publicly back in 2018, set a cultural baseline. He called himself “the opposite of a work-life balance role model” and told WIRED25 that he tries to take vacations but otherwise just works. That tone from the top shapes norms whether or not it is explicitly stated as company policy.
For those watching how the broader AI industry scales, this debate connects to a bigger question: does burnout risk undermine the very output these labs are racing to produce? Sustainable hiring and retention may matter as much as raw sprint speed, especially as the adoption of AI tools across enterprises creates long-term demand for experienced engineers, not just fast movers.
Our take
This story is easy to dismiss as Silicon Valley navel-gazing, but it has a practical angle for any business that relies on AI talent or partners with AI vendors. If the people building these models are burning out, that affects delivery timelines, model quality, and the pace of feature releases that downstream businesses depend on.
The “996” schedule signals something worth watching: as competition tightens, the labs most willing to push workers hardest will likely see shorter cycles but higher attrition. That trade-off never ends well long-term. For agencies and teams that use AI tools in production, it is worth having a plan that does not depend on any single model or vendor moving at a specific pace. Diversify your stack and avoid deep lock-in.
If you are thinking about how to integrate AI workflows without burning out your own team, our AI integration work focuses on sustainable adoption rather than chasing every new release.
What to do about it
- Audit which of your workflows are tightly coupled to a single AI vendor’s release schedule and identify alternatives.
- Set a review cadence (monthly, not daily) for evaluating new AI model releases rather than reacting to every announcement.
- Build internal documentation so your team’s AI knowledge is shared, not siloed in one person who cannot take a break.
- Watch attrition signals at the labs you depend on. High turnover at an AI provider is an early warning for product instability.
The labs racing hardest right now may not be the most reliable long-term partners. Pace matters.
Frequently asked questions
Do OpenAI and Anthropic employees get paid vacation?
Both companies have not publicly disclosed their specific PTO policies, and neither responded to interview requests on the topic. A 2025 Alliant survey found tech companies offer an average of 19.8 paid days off, and US Bureau of Labor Statistics data shows 95% of information workers have access to PTO.
What is the 996 work schedule in AI?
996 refers to working from 9am to 9pm, six days a week. Some software engineers in competitive tech and AI environments have adopted this schedule, sometimes alongside other lifestyle changes like cutting out alcohol, to maintain productivity.
What did Sam Altman say about work-life balance?
In 2018, Altman described himself as 'the opposite of a work-life balance role model.' At WIRED25 he said he tries to take vacations but otherwise just works.
Why do AI workers feel they cannot take vacations?
The fast pace of AI development means new model releases, benchmarks, and announcements appear constantly. Workers have described keeping up as 'drinking from a fire hose that keeps getting bigger,' creating pressure to stay connected even during planned time off.
