Sign in

Blog › Hacker News

Image: viarami on Pixabay
Hacker News · 3 min read

Should All AI-Generated Code Be Open Source? A Developer's Case

Published Nov 16, 2026

A developer argues that code produced by AI should be released as open source or under CC0. The claim touches training data, copyright law and a quiet shift in how software gets written.

A short essay by developer Jakub Jankiewicz is circulating in open-source circles with a blunt thesis: any code generated with the help of AI should be released under an open-source license or dedicated to the public domain with Creative Commons Zero (CC0). He calls it a hot take. It is also a compact summary of arguments that lawyers, maintainers and tool vendors have been circling for several years.

The training data argument

The first claim is about origins. Today's coding assistants learned from enormous amounts of publicly available code, much of it published under open-source licenses in the decades since Richard Stallman launched the GNU Project and the free software movement. They also learned from Stack Overflow and its sister sites, whose content carries Creative Commons licenses that require attribution.

According to the essay, scraping that material without permission or credit and then selling the output as closed, proprietary code amounts to license laundering. The underlying point is one many developers already feel: a commons built on reciprocity should not be converted into private property by a step that strips away every trace of where the knowledge came from.

This tension is not new. Lawsuits against code-generation tools have already tested whether training on licensed repositories respects the terms attached to them, and the question is far from settled in court.

Can machine-written code be owned at all?

The second argument is legal. Copyright, in many jurisdictions, protects works created by humans. The essay points to the well-known Monkey Selfie dispute, in which a photograph taken by an animal was found not to be protectable, and notes that the U.S. Copyright Office has drawn on the same logic when addressing AI output. Wikimedia Commons applies a similar rule in practice, requiring purely AI-generated images to be marked as public domain.

If that reasoning extends to software, then code written entirely by a machine may not be copyrightable in the first place, which makes a restrictive license on it a shaky foundation.

The picture is not uniform worldwide. The United Kingdom, for example, grants limited protection to computer-generated works. Because of those differences, the author suggests that developers and companies should make the choice explicit: release AI-generated code under an open-source license, or dedicate it to the public domain through CC0, so that its status is clear regardless of where it is used.

Where the line falls

The essay acknowledges that real projects rarely sit at either extreme. Most code is a mix of human and machine contributions. The author's concern is with the growing habit of accepting generated output without reviewing it, let alone adapting it. When a person's contribution is little more than a prompt, the argument goes, the result should be treated as open.

That is the fuzziest part of the proposal, and the part that matters most in practice. There is no agreed threshold for how much human editing turns generated code into an authored work. Projects, employers and courts will have to decide, and until they do, the boundary stays a judgment call.

Why it matters for developers

The practical stakes are concrete:

  • Licensing clarity. Teams shipping products with assistant-written code may find their ownership claims weaker than assumed.
  • Contribution policies. Open-source maintainers are already writing rules about AI-assisted pull requests, and provenance and licensing are central to those debates.
  • Compliance risk. Companies that treat generated code as fully proprietary could face questions from customers and auditors.

The proposal is unlikely to be adopted by vendors voluntarily, and it does not resolve the harder legal questions about training itself. Still, it frames a useful test: if a tool is built on the commons, what does it owe back?

What to watch next are court rulings on training data, further guidance from copyright offices, and whether major open-source foundations adopt explicit policies on AI-generated contributions. Those decisions will shape whether the essay's position stays a provocation or becomes a norm.

Original reporting: jakub.jankiewicz.org

Discussion

Loading comments …

The Repocharts newsletter

Once a week: the biggest repo clashes, the fastest risers and the AI and tech stories that matter. Free, no spam, unsubscribe with one click.

We send a confirmation link first (double opt-in). See our privacy policy.