How Many AI Apps Are on Google Play? What the Data Shows

The best available independent count is 36,344 apps with "AI" standing alone in the app name, found in an analysis of 3,037,501 public Google Play listings published by Testers Community in August 2026. Of those, 16,284 were released in 2026 alone. Google does not publish an official count of AI apps, so every number you will find comes from third-party listing databases, and the definition of "AI app" changes the answer.

Here is what that crowd actually means for you. The count is measured by app names, not by what the code does, so it is a floor and a marketing signal at the same time. It does not change any Google Play rule you have to clear: a new personal developer account still needs 12 testers opted in continuously for 14 days before production access, and how the app was built never appears in that requirement. What does change is the company you keep on the shelf, and one specific policy: if your app generates content with AI, Google's AI-Generated Content policy applies to it, with concrete in-app obligations.

This article covers what the independent data actually measured, what Google's own documentation requires from AI-built apps, and what our records from 1,500+ closed-testing campaigns show about the test itself.

What does the count of AI apps on Google Play actually measure?

Every published figure for "AI apps on Google Play" comes from scraping public listings, because Google publishes no such statistic. The most recent full analysis, by Testers Community, counts an app as AI if the standalone word "AI" appears in its name, which excludes false matches like "Air" or "Hair." Across 3,037,501 listings observed between 27 July and 25 August 2026, that method found 36,344 AI-named apps.

Three honest limits on that number, stated before you use it:

• Name-based counting undercounts real AI use. An app that summarizes your documents or drafts your emails may use a language model and never say "AI" in its name. Google's own policy documentation describes exactly this class of app as limited-scope AI apps.
• It overcounts "AI" as a marketing label. The same analysis found that naming an AI app no longer differentiates it: in 2026, roughly 1 in every 31 new apps carried "AI" in its name, against 1 in 350 in 2022. When the word describes a third of the shelf, it identifies a category rather than a feature.
• It is a snapshot of surviving listings, not of every app ever uploaded. Removed and suspended apps drop out.

So the defensible answer to "how many AI apps are on Google Play" is: about 36,000 by the name-based method, more by any functional definition, and no official number exists.

What does the independent data show about AI app launches?

Testers Community's August 2026 analysis is the most complete public dataset on the subject, and its most useful findings are not the raw count. All figures in the table below are theirs, from their analysis of 3,037,501 Google Play listings, not Google's statistics and not ours.

Three readings that matter more than the headline count. First, the median AI app and the median non-AI app land within about forty downloads of each other, which means "AI" in the name is not a distribution advantage. Second, the developer concentration figure says this is a long tail of small builders, not a few factories, so a new AI app is competing against a crowd of solo developers rather than a handful of studios. Third, more than a third of AI-named apps never reach 100 downloads, which is a distribution problem no naming decision can fix.

The same analysis found AI apps concentrate in crowded categories: Productivity at 7.1% of new releases, Entertainment at 6.0%, Tools at 3.3%, Business at 2.2%. If you launched a productivity app called "AI something" in 2026, the name placed you on the busiest shelf in the store.

Does the AI app flood change what Google requires from you?

No. The closed testing requirement is written against developer accounts, not against code. Google's documentation states that developers with personal accounts created after November 13, 2023 "must run a closed test for their app with a minimum of 12 testers who have been opted in continuously for at least 14 days" before they can apply for production access. There is no separate track, threshold, or exemption for apps built with AI tools, and no version of the rule that scales with category competition.

Nothing in the requirement text mentions how the app was produced, which framework shipped the build, or how many competing apps share your category. An app assembled in an afternoon from a builder prompt and a native app written over two years face the identical mechanical gate: 12 testers, opted in, 14 continuous days, then the production access application on the Play Console dashboard.

What does change under load is how your application reads. When you apply for production access, Google asks you to describe tester engagement and summarize the feedback you received, and it may require continued testing when "fewer than 12 opted-in testers or insufficient tester engagement during the testing period" is what your answers show. Those two failure conditions are the only documented reasons Google names, and both are about your test, not about the store around you.

In a category where most launches get no rating and under a hundred installs, the closed test is the only structured evidence of real usage your account will ever present to review. That evidence, not the shelf statistics above, is what the production access decision rests on. The 36,344 apps in the count never appear in your file; what appears is whether twelve real people used your app for two weeks and what you can say about it.

What does Google's AI-Generated Content policy require?

Here is the one Google rule that genuinely targets AI apps, and it is a content rule, not a build-tool rule. The AI-Generated Content policy applies to apps where generative AI output is a feature: text-to-text chatbots where the chatbot is a central feature, text-to-image and voice-to-image generators, and apps that create voice or video recordings of real individuals. For those apps, Google requires two concrete things:

• Apps "must comply with existing Google Play Developer Policies," including preventing the generation of restricted content such as child sexual abuse material and content that enables deceptive behavior.
• Apps "must contain in-app user reporting or flagging features that allow users to report or flag offensive content to developers without needing to exit the app."

The reporting requirement is the one vibe-coded apps most often miss, because a chat bolted onto an app during a build session has no report button unless someone adds one deliberately. The policy's scope documentation also matters for what it excludes: apps that merely host AI-generated content, apps whose only AI feature is summarizing non-AI content, and productivity features like AI-suggested email drafts are not covered at this time. If your app calls a model to generate new content from a user prompt, assume the policy applies; if it only summarizes or autofills inside an existing feature, it likely does not.

Google also stated its enforcement posture in August 2026. In an Android Developers Blog post dated 25 August 2026, its Trust & Safety team described reviewing thousands of generative AI apps across their lifecycle for controls against non-consensual intimate imagery, and stated that apps suspended or removed for generating or monetizing harmful content are blocked from monetization and advertising across Google's platforms. The practical reading: generative features are re-tested after launch, not once at submission, so a report flag you skip in version 1.0 becomes an enforcement problem in version 1.3.

Google's best-practices guidance for AI-generated content points developers at input and output filters and classifiers from model providers, and at safety-testing references including the Secure AI Framework and the OWASP generative AI red-teaming guide. None of this is a closed-testing requirement, but all of it belongs in the same QA pass you run during your 14 days.

What do we see in closed-testing campaigns for AI-built apps?

Based on our analysis of 1,500+ closed-testing campaigns run on OnTesters, the store-crowd question and the testing question resolve differently. Our records cover the closed test itself: physical devices (Samsung, Pixel, Xiaomi, OnePlus), Android 11 through 15, testers in more than 80 countries, matched typically in under 24 hours. They are our campaign records, not Google's data, and they say nothing about post-launch downloads.

What they do show is that AI-built apps fail the test for the same operational reasons as any other app, and the build tool is never one of them. The recurring patterns from our records: the app ships to testers as a web view of a development URL that dies when the builder session expires; the account is still a personal account created weeks earlier under the assumption that the store review would be quick; or the generative feature is present but nobody on the team wrote down what testers were supposed to exercise, so the engagement section of the production access questionnaire gets filled with a sentence instead of a summary.

The campaigns that run cleanly share one habit: the AI feature is treated as a feature with test cases, not as the reason the app exists. Testers are told what to prompt, what output to expect, and what to do when the model returns something wrong, and the developer keeps the answers. That is what the engagement questions in the production access application are looking for, and it is available to a two-day-old Replit project exactly as much as to a six-month native build.

What should you do before launching into that crowd?

1. Run the 14 days properly, not nominally. Twelve testers, opted in continuously, on the track that matches your account type. The clock and the count are the two things Google checks mechanically before you can even apply.
2. Write down what testers did. The production access application asks what features testers used and what feedback you received. Campaigns that kept a simple log fill that in; campaigns that did not get sent back to keep testing.
3. Ship the report flag if you generate content. In-app reporting or flagging, per the AI-Generated Content policy, before the closed test starts, so testers can exercise it and you can cite it.
4. Declare what your app sends out. If user prompts go to an AI service, that transmission belongs in your Data Safety declaration.
5. Meet the mechanical requirements early. An Android App Bundle and a target API level that meets the 2026 requirement are checked before anyone reads your feedback summary.
6. Do not expect the name to do work. The independent data is consistent on this point: with 1 in 31 new apps carrying "AI" in the name and median downloads barely separated from the pack, the pre-launch test is the only part of the launch you fully control.

Frequently asked questions

How many AI apps are on Google Play?

About 36,344 apps had the standalone word "AI" in their name as of August 2026, out of 3,037,501 public listings analyzed by Testers Community. No official Google figure exists, and any functional definition of "AI app" (apps using models without saying so in the name) produces a larger number.

Does Google publish an official count of AI apps?

No. Google publishes no statistic counting AI apps on Play. Every public figure comes from third-party listing analyses, which is why the method, such as name-based counting, always needs to be read alongside the number.

Do AI apps get more downloads than other apps?

Not by the available evidence. The median AI-named app recorded 345 lifetime downloads against 304 for the median app without AI in the name, and more than a third of AI-named apps never reach 100 downloads (Testers Community analysis, August 2026). Those are third-party listing figures, not Google data.

Will Google reject my app because it was built with AI?

No rule does that. The closed testing requirement applies to accounts created after November 13, 2023 and says nothing about how the app was written. What gets apps sent back for more testing is fewer than 12 opted-in testers or insufficient tester engagement, which are documented reasons on Google's production access help page.

Does Google's AI-Generated Content policy apply to my app?

It applies if your app generates content from prompts, such as a chatbot that is a central feature, an image generator, or an app creating voice or video of real people. It does not currently cover apps that only host AI content, apps whose only AI feature is summarization, or productivity features like suggested email drafts. Check the policy scope documentation against what your app actually does.

Do I still need 12 testers if my app was built with Replit, Lovable or Google AI Studio?

Yes. The requirement attaches to your developer account, not to the tool that produced the build. Our separate guide covers why AI-built apps still need 12 testers for 14 days, and our Replit publishing walkthrough shows where the clock starts in practice.

Is it too late to launch an AI app on Google Play?

No, but the name carries no weight anymore. When roughly 1 in 31 new apps says AI and the median AI app sees a few hundred installs, differentiation comes from the product and from a documented pre-launch test, not from the label. That is an interpretation of the third-party listing data above, not a Google statement.

Sources

• Testing requirements to publish on Google Play (Play Console Help) - the 12-tester, 14-day requirement for personal accounts created after November 13, 2023, the production access questionnaire, and the documented reasons for required continued testing.
• AI-Generated Content policy (Google Play Developer Policy Center) - restricted content obligations and the in-app reporting or flagging requirement for generative AI apps.
• Understanding Google Play's AI-Generated Content policy (Play Console Help) - which app types the policy covers and the limited-scope apps it currently excludes.
• Best Practices to Safeguard AI-Generated Content (Play Console Help) - filters, classifiers, and safety-testing references for generative features.
• Ensuring Safety in the Generative AI Ecosystem (Android Developers Blog, 25 August 2026) - Google's stated lifecycle testing and monetization enforcement posture for generative AI apps.
• How Many AI Apps Are on Google Play? The Real Number (Testers Community, 25 August 2026) - the independent listing analysis of 3,037,501 apps that every count in this article is attributed to.
• OnTesters campaign records, 1,500+ closed-testing campaigns on physical devices (Samsung, Pixel, Xiaomi, OnePlus), Android 11-15, 80+ countries. Our own records, not Google statistics, covering the closed-test period only.

How to read the numbers in this article. The store counts (36,344 AI-named apps, 16,284 releases in 2026, the 3.26% share, the 345 versus 304 median downloads, the 82.6% without a rating, the 28,189 developers) are all from Testers Community's independent analysis of public Google Play listings observed between 27 July and 25 August 2026, and are labeled as theirs wherever they appear. They are not Google statistics and they are not ours. The tester rules, the questionnaire contents, and the reasons for continued testing are quoted from Google's Play Console Help documentation and linked next to each claim. The observations about how AI-built campaigns run in practice come from OnTesters' own closed-testing campaign records and are labeled as such. Where this article draws a conclusion that no source states directly, such as what the crowd size implies for an individual launch, it is marked as interpretation rather than policy.