# Asahi Linux shifts to GPLv2+ licenses to prevent AI code lifting

> **Key takeaway:** Asahi Linux is moving its repositories from MIT to GPLv2 or later to protect unreleased code from being ingested and rewritten by AI models.

**Published:** 2026-11-12T13:52:53+00:00 
**Source:** It's FOSS Community News 
**Category:** distro-news 
**Canonical URL:** https://fosswire.org/news/asahi-linux-shifts-to-gplv2-licenses-to-prevent-ai-code-lift.html 

## Summary
The Asahi Linux project, which focuses on bringing Linux to Apple Silicon hardware, has announced a significant change to its licensing model. All project repositories are transitioning from the permissive MIT license to the GNU General Public License version 2 or later.

This strategic shift is a response to developers observing instances where unreleased code was being extracted and subsequently rewritten by AI tools to obscure its original source. By adopting a copyleft license, the project aims to ensure that derivative works remain within the open-source ecosystem and to prevent the unauthorized use of their engineering efforts in AI-generated content.

## Licensing shift and project motivations
- **License Transition:** The project is moving from the permissive MIT license to the more restrictive GPLv2 or later for all repositories.
- **AI Mitigation:** The change specifically targets the practice of 'lifting' unreleased code and using AI to rewrite it, which hides the original provenance of the software.
- **Hardware Context:** This development affects the software stack specifically optimized for Apple Silicon Mac architectures.
- **Legal Strategy:** The move utilizes copyleft requirements to ensure that code improvements and derivatives are shared back under compatible licenses.

## Why it matters
This move highlights a growing tension between open-source development and generative AI training practices. For the Linux ecosystem, it sets a precedent for how specialized hardware projects might use copyleft licensing to protect intellectual property from being absorbed into closed-loop AI models.