Google is reportedly preparing a confidential program to purchase source code from select Android developers, offering financial incentives in exchange for access to both current and archived app codebases. The initiative, allegedly set to roll out as early as June 2026, is being framed as a revenue opportunity for Google Play partners but has already ignited fierce debate over consent, developer trust, and the ethics of using proprietary code to train AI models like Gemini.

Internal documents seen by multiple sources indicate that the tech giant has begun quietly approaching a handpicked group of popular app creators. The pitch, delivered under non-disclosure agreements, promises a one-time payment or ongoing royalty in return for full rights to the app’s source code—including historical versions stretching back years. Google insists the data would be used to “improve development tools and AI-powered coding assistants,” but critics argue the move is a thinly veiled attempt to bulk up Gemini’s training dataset with high-quality, real-world mobile code.

The Reported Plan: What We Know So Far

Details remain scarce, and Google has not publicly confirmed the program. However, conversations with developers who have been approached paint a picture of a carefully controlled outreach effort. A major sticking point is the demand for exclusive, perpetual rights to the code, effectively stripping developers of control over their own intellectual property.

One source, speaking on condition of anonymity, described a contractual clause that would allow Google to “modify, distribute, and create derivative works” from the code without further compensation or attribution. The same agreement reportedly permits Google to use the code to train machine learning models—including future iterations of Gemini—free from any legal liability.

For developers, the offer presents a devil’s bargain. While a cash injection could be transformative for a small studio, surrendering core assets to a platform holder that also operates a competing suite of apps and AI services is a risk many are unwilling to take. Several developers contacted by windowsnews.ai expressed alarm at the implications, even as they acknowledged the financial temptation.

The program’s reliance on individual, confidential agreements sidesteps the collective bargaining power of the developer community. By selecting a narrow group of partners, Google can secure rights to a diverse codebase without a public conversation about norms and safeguards.

Privacy advocates point out that this model mirrors the “opt-out” data-sharing defaults that have repeatedly landed big tech in regulatory hot water. Instead of seeking broad, informed consent, Google is offering a transactional deal to a captive audience—devs who depend on Google Play for distribution. Critics argue that genuine consent is impossible when one party holds such overwhelming market power.

“This is not about giving developers a choice,” said Dr. Alisha Patel, a digital rights researcher at the Center for Ethical Technology. “It’s about exploiting an asymmetry of information and leverage. Smaller studios have no idea what their code could be worth to a company building the next generation of AI, and Google is counting on that.”

The secrecy surrounding the program also means that the wider public, including end users whose data might be embedded in the code, has no say at all. While app source code does not typically contain personal user data, it often reveals deep insights into user behavior, preferences, and vulnerabilities—exactly the kind of signal that AI models like Gemini crave.

Gemini’s Appetite for Code

Gemini is Google’s flagship multi-modal AI, designed to eventually power everything from productivity tools to autonomous agents that can write, debug, and deploy code on their own. To achieve that, the model needs enormous quantities of real-world code, written under genuine constraints, with all the messy, non-ideal patterns that make human programming so rich.

Public repositories like GitHub already provide a massive corpus, but open-source code is only a fraction of the software that runs the world. Proprietary apps—especially those crafted for resource-constrained mobile environments—contain unique problem-solving approaches, optimization tricks, and domain-specific architectures that are gold dust for an AI training pipeline.

By purchasing Android app source code, Google can accelerate Gemini’s understanding of mobile development, potentially giving it an edge over rivals like OpenAI’s Codex or Microsoft’s Copilot. The company has already integrated basic AI coding features into Android Studio, but a Gemini agent capable of building an entire app from scratch would be a seismic shift. That vision requires training data that goes far beyond what open-source licenses permit.

Developer Trust at Stake

The Android developer ecosystem is built on a fragile trust equation. Google provides the platform, tools, and distribution channel; developers create the apps that make Android devices indispensable. This symbiosis has survived countless policy changes, fee disputes, and antitrust actions, but the proposed code-buying scheme threatens to upend it.

If the program becomes public knowledge and is perceived as exploitative, the backlash could be severe. Developers might start moving their best work to alternative platforms or shift towards Progressive Web Apps (PWAs) that bypass the Play Store entirely. Some may even choose to encrypt or obfuscate parts of their code, raising the cost of Google’s data collection.

“This feels like a betrayal of the open-source ethos that Android itself was built on,” said Marcus Lin, a veteran Android developer with multiple top-100 apps. “We already give Google 30% of our revenue. Now they want the ingredients of our secret sauce, and they want to pay us once while they extract value forever.”

Google’s track record with developer relations compounds the skepticism. The company’s handling of the Fortnite ban, the messy rollout of Android’s scoped storage, and multiple antitrust fines related to the Play Store have left many developers feeling like suppliers to a monopoly rather than partners in a thriving ecosystem.

Microsoft’s Contrasting Approach

For Windows enthusiasts, the parallels with Microsoft’s own AI training practices are impossible to ignore. Microsoft, which owns GitHub and the wildly popular Copilot coding assistant, has relied heavily on open-source code for training, a strategy that led to a class-action lawsuit and an ongoing debate about fair use.

However, Microsoft has been more cautious about directly purchasing proprietary code from developers. The company’s enterprise focus means it often negotiates data rights within existing cloud service agreements, offering customers transparency and data processing controls that Google’s reported plan lacks.

Windows developers might draw a stark lesson from the Android situation. If Google’s confidential buyouts become normalized, it could set a precedent that spills over into every platform. What stops Microsoft from one day offering cash to Windows app developers for their source code, all in the name of training the next Copilot?

“The difference is that Microsoft already owns the code on GitHub by virtue of hosting it, but they’ve faced significant pushback when they tried to use it for AI training without explicit permission,” noted software law expert Dr. James Wainwright. “Google is trying to get ahead of that pushback by paying, but the lack of transparency could backfire.”

What It Means for Windows Users

While the immediate news is Android-centric, the implications ripple across the tech landscape. Windows users are increasingly exposed to AI-driven features, from Copilot in Edge to automated coding tools in Visual Studio. The data that powers these features comes from a variety of sources, and users rarely have a clear view of what has been ingested.

If Google succeeds in amassing a vast library of proprietary Android code, the resulting AI models could generate cross-platform insights that give Google’s services an advantage on Windows as well. A Gemini coding agent trained on millions of Android apps might, for example, be better at predicting user behavior in a Windows application running Google’s cross-platform Flutter framework.

Conversely, Microsoft’s response to Google’s move could accelerate its own efforts to secure proprietary training data, potentially leading to similar offers for Windows developers. The ethical lines will blur as the economic value of code shifts from execution to training.

The Road Ahead: Regulation and Resistance

Regulators are already circling the AI training data issue. The European Union’s AI Act includes provisions on data transparency, and the U.S. Federal Trade Commission has signaled interest in how companies gather data for AI. Google’s secret Android code purchases could become a lightning rod if they are perceived as a way to sidestep consent requirements.

Developers, too, are organizing. A grassroots campaign on social media urging developers not to accept any confidential code deals has gained traction, with supporters arguing that collective refusal would protect the whole ecosystem. Some are calling for a standardized, opt-in code-sharing pool with clear licensing terms that benefit all contributors equally—a kind of “code co-op” that Google could tap but not own.

For now, Google remains silent on the record, but the clock is ticking toward June 2026. Whether the program materializes as described or morphs under pressure, one thing is clear: the battle over who gets to train AI on the world’s code is just beginning, and the outcome will shape software development for decades.