Legal Isn't the Same as Legitimate: AI Rewriting Code and the Erosion of Copyleft

1 views

A legal-and-ethical long read argues that AI "reimplementing" open source code may legally sidestep Copyleft licenses while eroding their spiritual foundation.

The New Loophole Copyleft Faces

Copyleft licenses like the GPL are ingeniously designed: if you use my code, your derivative work must also be open source, spreading freedom like a virus. The author points out that AI has torn open a gap: have a model "learn" a piece of GPL code, then generate a functionally equivalent but differently expressed implementation, which literally may not constitute a "derivative work," so Copyleft's contagion clause gets bypassed. The code's functionality is freeloaded, the obligation to give back to the community is laundered away, the license's legal shell remains, but its soul is gutted.

The Fissure Between Legal and Legitimate

The article's title names the core: this laundering "may be legal," but it clearly isn't "legitimate." The collaborative ethic the open source movement built over decades rests on the tacit understanding that "you take, you give back," and AI rewriting offers a shortcut that's formally compliant but substantively a betrayal. The lag of the law is laid bare here: copyright law protects expression, not function, and AI can precisely preserve function while replacing expression. This debate has no ready answer, but the question it poses is sharp: when technology makes "circumventing the spirit of a license" trivially easy, can the trust structure the open source community runs on still hold? This is a question hanging over the entire free software movement.

via: Hacker News