Google has been quietly contacting Android developers with a proposal to purchase access to their source code. According to 404 Media, the firm emailed a select group of Google Play developers, inviting them to participate in what it describes as a “confidential content offer pilot.”
The message presents the deal as a revenue stream, stating that developers can “get paid for sharing the code powering your apps, as well as your archived projects.” Google assures that developers keep their intellectual‑property rights and that the license granted is non‑exclusive.
So what is Google after?
The report notes that the email never mentions artificial intelligence, but a hidden link leads to a page titled “partnerships to improve our AI products.” That page openly says Google is paying for “non‑public content in a range of media formats” to enhance its AI models.
The connection is clear. Google’s Gemini excels at image and text generation but lags behind in AI‑driven coding tools, while Anthropic’s Claude Code has achieved a valuation surpassing OpenAI’s. OpenAI has also released its own Codex app aimed at developers, and at the recent Google I/O the company showcased the Antigravity 2.0 IDE capable of generating full apps.
It appears Google wants to train its AI on real‑world code to boost its coding capabilities and compete with Claude Code and ChatGPT’s Codex. Purchasing actual developer code offers a shortcut to narrowing that gap.
Is anything amiss?
While the long‑term effects could be harmful to developers, the approach isn’t inherently unethical. It’s arguably better than training AI on massive corpora of books and online material without permission, a practice many AI firms have employed.
Developers retain their IP, receive a non‑exclusive license, and get paid. However, the lack of transparency in the email is notable. Pitching an AI data‑gathering program as a simple “revenue opportunity” without mentioning AI at all suggests Google hopes developers won’t probe further.