--- title: "Duolingo CEO Reverses AI-First Memo in 2026 | Metaintro" canonical: "https://www.metaintro.com/blog/duolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026" language: "en" author: "drashtigarach" published: "2026-05-13T17:49:46.000Z" modified: "2026-10-02T19:27:05.746Z" --- [Back to Blog](/blog) [News](/blog/tag/news)[Information](/blog/tag/information)[AI](/blog/tag/ai) # Duolingo CEO Reverses AI-First Memo in 2026: "We've Never Laid Off" Duolingo CEO Luis von Ahn walked back the 2025 AI-first memo in May 2026, scrapping the AI usage performance rule as full-time hiring keeps growing in 2026. [![Drashti Garach](https://cdn.metaintro.com/rs:fill:40:40/q:72/plain/images/5719d740-e510-42bc-8017-e040d145f35f_1766029465094.png)Drashti Garach @DrashtiGarach](/blog/author/drashtigarach) [May 13, 2026](/blog/archive/2026/05)Updated October 2, 202612 min read ![Duolingo CEO Reverses AI-First Memo in 2026: "We've Never Laid Off"](https://cdn.metaintro.com/rs:fill:1200:675/q:78/plain/images/kai.bGx1nFep.png) [https://x.com/intent/tweet?text=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026](https://x.com/intent/tweet?text=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026&title=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026&title=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22)[mailto:?subject=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026](mailto:?subject=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026) Luis von Ahn does not want to be the face of AI-first job cuts anymore. On a May 13, 2026 episode of the Rapid Response podcast with host Bob Safian, the Duolingo CEO admitted the company "backtracked" on the rule that pushed every employee to be evaluated on AI usage, because workers kept asking whether they were being told to use the tools for AI's sake. He also pointed out a fact that got lost in the original 2025 backlash. Duolingo has never laid off a single full-time employee, and headcount kept growing the year of the memo that scared the internet. The reversal matters because plenty of CEOs are still copying that 2025 playbook, and many of them are pairing AI-first language with actual workforce cuts that the [Sinch enterprise rollback survey](https://www.metaintro.com/blog/74-percent-enterprises-rolled-back-ai-customer-agents-sinch-2026) and the [NBER survey on the AI productivity mirage](https://www.metaintro.com/blog/executives-ai-productivity-mirage-job-security-nber-survey-2026) have already flagged as premature. For job seekers reading job descriptions and earnings calls, Duolingo is now the most useful counter-example in the market. ## What did von Ahn actually take back? Von Ahn did not abandon AI. He abandoned the metric. On the podcast, he described what changed inside the company. "At the end, we backtracked, and we said, 'No. Look, the most important thing in your performance is that you are doing whatever your job is as well as possible. A lot of times AI can help you with that. But if it can't, I'm not going to force you to do that.'" The trigger was internal feedback. Employees told leadership that the requirement was producing weird incentives, where teams reached for an AI step they did not need just to show up on the dashboard. "It felt like rather than being held accountable for the actual outcome, we're trying to just push something that in some cases did not fit," he said. The retracted rule is the same one that drew most of the heat in the original AI-first memo on April 28, 2025. Performance reviews tied to AI usage struck a nerve because that is the exact mechanism workers see in their own offices, and it is the same mechanism flagged in the [Amazon AI usage targets story](https://www.metaintro.com/blog/amazon-ai-usage-targets-inflated) where managers inflated tool adoption numbers. Von Ahn is signaling that even at a company that genuinely leans into automation, forcing the metric backfires, and that aligns with what the [executives AI productivity mirage NBER survey](https://www.metaintro.com/blog/executives-ai-productivity-mirage-job-security-nber-survey-2026) has been arguing all year. The lesson for managers is that input metrics measure compliance, not capability. The lesson for workers is that you can push back on a bad mandate and sometimes the CEO actually listens. ## Has Duolingo actually avoided layoffs? Yes, and that is the part that complicates the AI-doom narrative most. Von Ahn has repeated across multiple interviews that Duolingo has never laid off a full-time employee since its founding in 2009, and that headcount kept growing the year of the AI-first memo. Reporting from Fortune and CNBC confirms the company added full-time roles after the April 2025 announcement, even while pulling back on contractor work. That stands in sharp contrast to the [78,557 tech layoffs Q1 2026 tally](https://www.metaintro.com/blog/78557-tech-layoffs-q1-2026-ai-automation-workforce-cuts) and to companies like [Block under Jack Dorsey](https://www.metaintro.com/blog/block-jack-dorsey-ai-layoffs-excellence-strategy-2026), [Coinbase with its 700-person AI pivot](https://www.metaintro.com/blog/coinbase-700-layoffs-ai-pivot-2026), and [Freshworks cutting 500](https://www.metaintro.com/blog/freshworks-500-layoffs-ai-writes-half-code-tech-workers-2026), where AI talking points showed up the same week as cuts. Duolingo gives job seekers a useful test. When a CEO says "AI-first," ask whether the company is actually hiring or actually trimming. The [AI layoff myth piece](https://www.metaintro.com/blog/ai-layoff-myth-companies-using-ai-cover-job-cuts) and the [corporate euphemism decoder](https://www.metaintro.com/blog/ai-layoffs-corporate-euphemism-decoder-2026) both argue that AI is regularly used as cover, and Duolingo's headcount data is the rare counter-example. The [cover story on AI layoffs](https://www.metaintro.com/blog/ai-layoffs-cover-story-2026) adds more receipts. Contractors are a different conversation. Duolingo phased some of them out as it leaned on automation, and that decision is worth weighing if you are a freelance translator, illustrator, or audio producer. The trade-off Von Ahn is offering is concrete. Full-time, salaried employees keep their seats. Contract work shifts toward review, quality control, and edge-case judgment, the work that survives an Anthropic AI jobs report career planning read of the labor market. ## Why did the AI usage rule backfire inside the company? Workers were doing what the system asked, and the system was asking for the wrong thing. When AI use is a graded input rather than a means to an outcome, employees optimize for visible activity, not better work. Researchers have been calling out this pattern for months. The [Microsoft 2026 study on AI leadership](https://www.metaintro.com/blog/microsoft-2026-study-ai-leadership-not-workers) places the responsibility for integration on managers, not on individual contributors. The [AI productivity paradox roundup](https://www.metaintro.com/blog/ai-productivity-paradox-2026-companies-no-gains-economists-breakthrough) and the [90 percent of companies showing zero gains analysis](https://www.metaintro.com/blog/why-90-percent-companies-zero-ai-productivity-gains-job-seekers-2026) describe the same dynamic at scale. Mandates without judgment do not produce productivity. They produce [workslop](https://www.metaintro.com/blog/workslop-ai-productivity-workplace-crisis-2026), the lightly polished, plausible-but-empty output the [AI productivity trap article](https://www.metaintro.com/blog/ai-productivity-trap-workers-heavier-workloads-employers-raise-expectations-2026) describes as quietly raising expectations without improving real performance. Von Ahn's reversal is rare because most CEOs double down. The [Anthropic 10 percent performance divide research](https://www.metaintro.com/blog/anthropic-data-10-percent-performance-divide-ai-power-users-newcomers-2026) shows that AI gains cluster in a small group of skilled users, and forcing everyone to mimic them does not transfer the skill, it just dilutes the signal. The [workplace culture decides AI project success piece](https://www.metaintro.com/blog/workplace-culture-decides-ai-project-success-2026) makes the same point from a different angle, that integration is a culture problem before it is a tool problem. There is a downstream cost too. The [workers won't speak up on AI errors story](https://www.metaintro.com/blog/workers-wont-speak-up-ai-errors-workplace-silence-risk-2026) shows what happens when employees fear flagging the bad outputs the mandate is generating, and the [AI agents are not employees HBR research](https://www.metaintro.com/blog/ai-agents-not-employees-hbr-research-2026) frames why ownership cannot be outsourced to a model. ## What does von Ahn mean by AI "slop"? He means the same thing your favorite Reddit thread means. AI can produce a lot of competent-looking content, and a chunk of it is unusable. Duolingo's product depends on producing thousands of short stories per language to power lessons, and von Ahn has publicly conceded that generative tools are not yet reliable at that volume. Long-running reporting has tracked user complaints about awkward translations and bland writing inside the app after Duolingo accelerated AI-generated content. The slop problem is not a Duolingo problem. It is a category problem, and it has already shown up in the [AI fake citations in courts story](https://www.metaintro.com/blog/ai-fake-citations-courts-legal-careers-2026), in the [Hollywood writers retraining as Mercor AI trainers piece](https://www.metaintro.com/blog/hollywood-writers-mercor-ai-trainers), and in the [Swedish cafe AI agent experiment](https://www.metaintro.com/blog/ai-agent-swedish-cafe-experiment) where automation hit a quality wall fast. The [company's AI nightmares HBR piece](https://www.metaintro.com/blog/companys-ai-nightmares-hbr) catalogs more of the same. The takeaway for job seekers is concrete. Companies that need volume and care about quality will hire human judgment to filter, edit, and rewrite AI output. That is exactly the work the [layer AI skills on your existing career guide](https://www.metaintro.com/blog/layer-ai-skills-existing-career-without-starting-over) describes, and it lines up with the [AI biggest impact on minds not jobs analysis](https://www.metaintro.com/blog/ai-biggest-impact-minds-not-jobs-cognitive-workforce-2026), which argues that cognitive review work is the durable layer. Even von Ahn, who has every incentive to talk up automation, is admitting that creativity and polish still belong to the humans, and he is hiring accordingly. ## What does this mean for your job search in 2026? It means AI fluency still helps, but AI mandates are a yellow flag, not a green one. The Duolingo reversal is the clearest CEO-level admission this year that forcing AI down a workforce does not improve output. Job seekers should read it the same way the [CEOs fear losing jobs if AI bets fail piece](https://www.metaintro.com/blog/ceos-fear-losing-jobs-ai-bets-fail-2026) reads it, as a signal that the C-suite is recalibrating. Hiring is moving toward people who can show AI fluency where it actually compounds their work, not toward people who can game an AI usage dashboard. That matches the [Anthropic 30 billion raise and AI hiring trend](https://www.metaintro.com/blog/anthropic-30-billion-raise-900b-valuation-ai-hiring-2026), the [Google forward-deployed engineer push](https://www.metaintro.com/blog/google-forward-deployed-engineers-hiring-ai-2026), and the [Salesforce class of 2026 hiring 1,000 AI-native grads](https://www.metaintro.com/blog/salesforce-1000-ai-native-grads-class-of-2026). It also lines up with the [AI productivity pay gap data](https://www.metaintro.com/blog/ai-productivity-pay-gap-workers-wages-2026), which shows AI-skilled workers commanding measurable wage premiums. Practical moves now. Read the [how to show AI fluency on your resume guide](https://www.metaintro.com/blog/how-to-show-ai-fluency-on-your-resume-2026) before the next round of applications, study the Anthropic AI jobs report for career planning, and use the [11x growth in AI role searches breakdown](https://www.metaintro.com/blog/ai-role-searches-grown-11x-career-pivot-2026) to spot which titles are actually expanding. Cross-reference with the [2026 career skills to stay employable guide](https://www.metaintro.com/blog/2026-career-skills-stay-employable-constant-change) and the [workers using AI as a business coach piece](https://www.metaintro.com/blog/workers-using-ai-business-coach-career-advice-2026) for tactical resume and interview prep. When you interview, ask the hiring manager what the AI policy looks like in practice, and ask whether usage is tracked in performance reviews. If the answer sounds like Duolingo's old memo, treat it the way von Ahn now does, as a draft to be revised. If the answer sounds like the new memo, where AI is a tool and outcomes are the metric, you are probably looking at a workplace that will still be hiring humans in 2027. Here is the deeper read. Duolingo is the rare case where the headline (AI-first) and the labor data (zero layoffs, growing full-time headcount) point in opposite directions, and that gap is exactly what job seekers in 2026 should be learning to read. The mistake most candidates make right now is treating every AI mention in a job description, a CEO post, or an earnings call as the same signal. It is not. An AI-first memo paired with growing headcount and a public reversal on a bad metric is a company iterating in public. An AI-first memo paired with quarter-over-quarter workforce cuts and vague language about productivity is a company using AI as a rhetorical wrapper for cost reduction. The two look identical on the surface and behave nothing alike inside the building. When you are evaluating a role this year, slow down and check three things before you accept the framing the company is selling. Look at full-time headcount trend lines in the last four quarters. Look at whether contractor counts moved the same direction or the opposite. Look at whether the CEO has ever walked anything back publicly when employees pushed. Companies that pass those three checks tend to be the ones still hiring humans in twelve months. Companies that fail them tend to be the ones explaining away the next round of cuts. Duolingo is currently passing. That does not mean Duolingo is a guaranteed safe harbor forever, and it does not mean every Duolingo team is hiring. It means the company has provided one concrete data point that the job-seeker community can use to calibrate against the louder, scarier AI-first rhetoric coming from every other corner of the market this year. Use it as a benchmark, not as a destination. One more practical note for the application stack itself. When a recruiter asks how you use AI in your work, do not recite a tool list. Describe the outcome you produced and where AI sped it up, where AI got in the way, and where you chose to skip the tool entirely because the judgment call was yours. That answer maps to exactly the kind of worker von Ahn says he wants more of after the reversal, and it maps to the kind of worker most CEOs will eventually admit they want once the AI-first marketing burns off. It also makes you legible to hiring managers who are themselves still figuring out what the policy should be in their own building. ## People Also Asked ### Q: Did Duolingo lay off employees because of AI? A: No. CEO Luis von Ahn has stated that Duolingo has never laid off a single full-time employee since the company was founded in 2009, and that full-time headcount actually grew the year of the April 2025 AI-first memo. The company did reduce contractor roles as it leaned more on automation for content generation, which is a different category of work and a different employment relationship. ### Q: What did Luis von Ahn say on the Rapid Response podcast? A: Speaking with host Bob Safian on May 13, 2026, von Ahn said Duolingo backtracked on the rule that required all employees to be evaluated on AI usage, because workers asked whether the company wanted them to use AI just for AI's sake. He said the most important thing in performance is doing the job as well as possible, and AI should help where it helps, not be forced where it does not fit. ### Q: Is Duolingo still an AI-first company in 2026? A: Von Ahn says yes in strategy and no in mandate. He remains a believer in AI as a productivity tool and wants to hire more workers who use it well, but he removed the performance review requirement, acknowledged AI slop as a real product problem, and reframed the company's stance around outcomes rather than tool usage frequency. Future-proof your career. [Metaintro](https://www.metaintro.com) tracks AI-disrupted roles and surfaces the human-skill jobs growing alongside automation. ### Share this article [https://x.com/intent/tweet?text=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026](https://x.com/intent/tweet?text=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026&title=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026&title=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22)[mailto:?subject=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026](mailto:?subject=Duolingo%20CEO%20Reverses%20AI-First%20Memo%20in%202026%3A%20%22We've%20Never%20Laid%20Off%22&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fduolingo-ceo-walks-back-ai-first-memo-hiring-grows-2026) For job seekers ## Your next role is already posted. Say what you want next. We search every day and send real, open roles, with help getting ready for each. [Get Started Free](/signup)[Search matching jobs](/jobs/search) - MatchCompare live roles against your current evidence. - PositionTurn proof projects into role-specific applications. - ImproveUse market feedback to keep the skill plan current. [Return to navigation](#main-navigation)