Developer Migrates from GitHub to Forgejo Over AI Training Concerns

A developer who has used GitHub since 2012 has departed the platform, migrating private repositories to a self-hosted Forgejo instance. The move signals growing tension between the convenience of centralized version control and the desire for absolute data sovereignty.

Writing on log.ozgur.works, the author detailed the decision to move their workflow to a personal homelab and cancel their GitHub Copilot subscription. The primary drivers for the migration were perceived software instability and a fundamental distrust of how Big Tech handles privacy and security. Specifically, the author pointed to the use of user code to train artificial intelligence models as a critical breaking point. "Do not trust companies, especially big tech, when it comes to privacy or security," the author wrote. "Be suspicious."

The Push Toward Sovereignty

This migration reflects a broader movement within the tech community toward "de-googling" and the adoption of homelabbing. As large language models (LLMs) continue to be trained on vast swaths of open-source and user-submitted code, developers are increasingly questioning the terms of service of centralized platforms. The shift toward Forgejo—a community-driven, self-hosted git service—allows developers to maintain total control over their intellectual property and infrastructure, removing the intermediary between the coder and their data.

The Barrier of Vendor Lock-in

While individual developers can migrate their repositories with relative ease, the transition highlights a significant divide in the industry regarding vendor lock-in. Discussions on Hacker News following the announcement suggest that while individuals can seek sovereignty, many companies remain trapped by complex integrations. The primary obstacle is often the reliance on proprietary CI/CD pipelines that make moving to a different platform technically prohibitive.

One Hacker News user noted that teams who design their entire build process around a specific platform's CI become "brittle," eventually reaching a state where they can only execute portions of their build on that specific infrastructure. This dependency creates a systemic barrier that prevents organizations from leaving centralized providers, even when ethical or stability concerns arise.

Future Outlook

As AI training continues to evolve, the conflict over code ownership is expected to intensify. The industry remains divided on whether current copyright frameworks sufficiently protect developers from having their work ingested by LLMs. For now, the trend toward self-hosting serves as a practical hedge for those unwilling to accept the privacy trade-offs of the modern cloud ecosystem.