Google Is Quietly Buying Code From Play Store Developers to Train AI

Disclosure: Some links in this article are affiliate links. AI Maestro may earn a commission if you make a purchase, at no…

By AI Maestro June 3, 2026 3 min read
Google Is Quietly Buying Code From Play Store Developers to Train AI

For the makers and artists building the next generation of software, the landscape is shifting from a wild west of open data to a marketplace where your own creations carry a price tag. Google has quietly initiated a pilot programme to purchase direct access to the source code of Android applications hosted on the Play Store, a move designed to feed its proprietary AI coding models.

The initiative, confirmed by 404 Media, involves emails dispatched to app developers offering them a chance to “join a confidential content offer pilot.” The correspondence promises that participants can “generate additional revenue from your apps” by granting Google access to their codebases to “help improve Google’s developer tools and products.” Because the programme is described as confidential, the publication granted the recipient anonymity, noting fears of potential retaliation for exposing the arrangement.

The mechanics of the offer

The outreach explicitly invites developers to monetise not just their live applications but also their historical work. The email states: “Get paid for sharing the code powering your apps, as well as your archived projects.”

Crucially, the terms maintain the developer’s ownership. The message assures recipients that they will retain intellectual property rights and that the licence is non-exclusive. The pitch frames the transaction as a unique chance to “help transform tools and products, support the developer ecosystem, and unlock new revenue,” covering everything from active production code to prototypes and side projects that are no longer in use.

While the email text conspicuously omits any mention of artificial intelligence, a hyperlink within the message directs recipients to a page titled “partnerships to improve our AI products.” This landing page clarifies that Google is seeking to “pay for the delivery of non-public content in a range of media formats,” moving beyond the publicly scraped data that most AI firms currently rely upon.

Google frames this acquisition as a mission-driven endeavour. The company argues that paying for content allows it to “help individuals, helping businesses, [and] helping society at large,” citing examples such as aiding in the management of natural disasters and helping doctors detect diseases earlier. The rhetoric positions the training of AI tools as a “once-in-a-generation opportunity to help the world combat and manage natural disasters.”

A strategic pivot in the race

For Google, this shift represents a significant admission of vulnerability. The tech giant has trailed its rivals in the specific domain of code generation. Competitors like Anthropic have leveraged the success of Claude Code to achieve a valuation exceeding that of OpenAI, while Microsoft’s Copilot has secured widespread adoption.

The decision to solicit code directly from creators suggests that Google has struggled to construct a sufficiently robust coding AI using only the content it can scrape from the public web. It highlights a broader industry reality: companies are rapidly exhausting the easily accessible reservoir of training data. This approach mirrors Google’s previous, high-profile attempt to pay Reddit $60 million for access to its site for AI training, an initiative that ultimately yielded mixed results for the search giant.

“We are reaching out on behalf of the Google Partnerships team with an invitation for a select group of Google Play app developers to join a confidential content offer pilot.”

Key takeaways

  • Google is actively soliciting direct payment from Play Store developers to access their source code, marking a transition from public data scraping to private data purchasing.
  • The programme offers non-exclusive licences where developers retain 100% of their intellectual property rights, allowing them to monetise the code elsewhere.
  • This strategic move underscores the scarcity of high-quality, public-domain training data and Google’s need to catch up to competitors in the field of AI code generation.

Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.

Name
Scroll to Top