Addictive learning, on purpose.
There is a sentence inside Gizmo's Series A announcement that is worth reading twice. Petros Christodoulou, CEO, puts it this way: "We're fighting for better screen time." The word "fighting" is doing unusual work there. Most study-tool founders would reach for something softer — enabling, empowering, unlocking. Christodoulou reaches for a combative frame, because he is describing a combative position: the same engagement mechanics that a decade of public debate has taught us to treat as the enemy of learning are, in his framing, the payload. Not a side effect to apologize for. The mechanism.
In April 2026, with thirteen million users and twenty-two million dollars in fresh Series A funding — led by Shine Capital with Ada Ventures, GSV, and NFX among the participants — Gizmo has made the bet visible enough to sit with. The bet is not subtle. And the question it opens is the kind worth reading slowly: is "addictive learning" a category error, or a category that has not been properly invented yet?
What the inversion is claiming
To understand Gizmo's positioning, it helps to name what it is inverting. The dominant framing in edtech messaging since roughly 2018 has been a binary: screen time bad, learning good. Products were supposed to be the antidote to the phone — the responsible alternative, the one parents could feel calm about. The pitch was calibrated for the adult holding the credit card. It quietly accepted the premise that the engagement mechanics of consumer apps were, by definition, destructive.
Gizmo's framing refuses that premise. The company takes the four forces that consumer-app engineers have used to hold attention — progress signals, novelty, social comparison, variable reward — and points them at a flashcard. The claim is not that these mechanics are safe because they're aimed at study. The claim is more specific: that the mechanics are neutral carriers. What they're carrying is what's load-bearing.
This is not the same as gamification, which adds points and badges to an existing learning structure. Addictive learning re-engineers the product around variable-reward loops first and treats the study content as the payload running inside that architecture. The product is built the way a social feed is built — except the feed is flashcards.
The scale of adoption suggests the mechanics are working in at least one narrow sense: thirteen million users is a number that most study-tool companies do not reach. The product is sticky. Whether "sticky" and "learning" point at the same thing is the harder question — and one the field hasn't yet given a clean answer to.
The regulatory collision
What makes this moment genuinely interesting is the collision. Gizmo's Series A arrives inside a 2026 policy landscape that is, in several US states, actively working to ban the mechanics the product is built on.
The 2026 edtech backlash is not a single movement. It is two policy fronts arriving at once. The first is the anti-screentime expansion — seventeen or more US states considering bills to restrict tech use during school hours, and the KOSMA bill reaching to regulate school-issued devices. The second is California's Parents & Kids Safe AI Act, announced January 9, 2026, co-sponsored by OpenAI and Common Sense Media, which bans targeted advertising to minors, mandates safeguards against harmful content, and — most relevant here — targets the dark patterns that consumer apps use to hold a child's attention.
The regulatory text, as written, does not yet distinguish between addictive mechanics aimed at social feeds and addictive mechanics aimed at study. The word the regulation reaches for is not "feed" — it is "engagement mechanic." Gizmo's defense relies entirely on a distinction that the policy has not yet formalized.
This is not a verdict on the company. It is a noticing. The product category may not survive 2027 in its current framing — or the regulatory text may sharpen to make the distinction Gizmo's founders believe is real. Both are live possibilities. What is clear is that the company has positioned itself at exactly the point of maximum regulatory pressure.
The test the OECD is offering
The question underneath the noise is not regulatory; it is pedagogical. Does the engagement actually track with learning? The most useful instrument for asking that question in 2026 is the OECD Digital Education Outlook, published this year, which finds that metacognitive engagement — the ability to monitor and regulate one's own thinking — declines when generative AI does the cognitive work that the student was meant to do. OECD Digital Education Outlook 2026
The OECD finding is not about flashcard apps specifically. But it contains the sharpest version of the challenge addictive learning has to meet: engagement is not metacognitive engagement. A student who is genuinely absorbed — who is completing the next card, chasing the next streak, wanting to see the next score — is demonstrating behavioral engagement. That is real. It is not nothing. But the OECD finding points at a different layer: whether the student, while engaged, is also building the capacity to regulate their own learning. Variable-reward loops are, by design, exogenous regulation. The loop tells you when to engage, how long to persist, when the session is complete. The student's own metacognitive apparatus is not what's doing that work.
This is the question worth sitting with. It is not a condemnation — it is a design challenge. An addictive-learning product that also builds metacognitive capacity would be something genuinely new: it would have to point the variable-reward loop not just at correct answers but at the student's own process of arriving at them. The difference between "you got this one right, try the next" and "notice that you hesitated before answering — that hesitation is information" is the difference between behavioral engagement and metacognitive engagement. Building for the second while maintaining the engagement properties of the first is the architecture the category has not yet invented.
What "better screen time" would need to prove
Gizmo's Series A comes with a stated mission of helping a billion people access engaging education globally. That is a generous ambition, and the funding suggests serious investors believe it is achievable. What it would take to earn "addictive learning" as a name — rather than merely as a positioning — is a specific demonstration.
The demonstration would show that engagement on the platform is correlated with durable learning: not short-term retention on the immediate quiz, but transfer — the ability to use what was learned in a new context after weeks, not minutes. And it would show that sustained use does not erode the metacognitive capacity to learn without the scaffold. The second requirement is the harder one, because the business model runs on the scaffold. A student who no longer needs the app is a student who churns.
This is not a problem unique to Gizmo. It is the structural tension inside any habit-forming learning product. The best possible version of the category resolves it by designing the scaffold to gradually externalize its regulatory function — building the student's own capacity rather than substituting for it, until the app becomes optional rather than compulsory.
That product does not appear to exist yet at scale. The category is young, the evidence base is thin, and the OECD has just handed the field a better test. Whether the builders in this space — Gizmo among them — will pick it up is, genuinely, one of the more interesting bets in education right now. We are watching it with the curiosity it deserves, and none of the certainty it hasn't earned.
Sources
Wiki pages drawn from
concepts/addictive-learning— the 2026 design framing: variable-reward loops pointed at study content; the Gizmo CEO quote as canonical instance; distinction from gamification; open questions on metacognitive stress-test.entities/companies/gizmo— Cambridge founders, $22M Series A (Shine Capital, Ada Ventures, GSV, NFX), 13M users, "fighting for better screen time" positioning.topics/edtech-backlash-2026— 17+ US state bills, KOSMA expansion, California Parents & Kids Safe AI Act (OpenAI + Common Sense Media), regulatory text's current failure to distinguish engagement mechanics by destination.
External sources
- "Gizmo bags $22M in Series A funding as it focuses on making learning addictive" — EdTech Innovation Hub, April 16, 2026. https://www.edtechinnovationhub.com/news/gizmo-bags-22-million-in-series-a-funding-as-it-focuses-on-making-learning-addictive
- "Two Fronts Reshaping EdTech Policy" — Edtech Insiders (Ben Kornell), January 20, 2026. https://edtechinsiders.substack.com/p/the-two-fronts-reshaping-edtech-policy
- OECD Digital Education Outlook 2026 — OECD. https://www.oecd.org/education/oecd-digital-education-outlook-2026.htm