Development quote of the week [LWN.net]

"By Wednesday morning, Anthropic representatives had used a copyright takedown request to force the removal of more than 8,000 copies and adaptations of the raw Claude Code instructions—known as source code—that developers had shared on programming platform GitHub."
Because if there's one thing GenAI companies absolutely don't take lightly, it's copyright.

— Jens Ohlig commenting on a quote from a Wall Street Journal article.

to post comments

Getting the cat back into the bag

Posted Apr 2, 2026 10:51 UTC (Thu) by anselm (subscriber, #2796) [Link] (7 responses)

The other thing is that Claude Code is (according to Anthropic) 100% vibe-coded, and in the USA so far, copyright does not apply to LLM output in the first place. AFAIR, claiming copyright on stuff that you don't in fact have the copyright for is fraudulent, and doing so in a DMCA takedown notice (like Anthropic is purportedly doing) is perjury.

Getting the cat back into the bag

Posted Apr 2, 2026 10:56 UTC (Thu) by bluca (subscriber, #118303) [Link] (6 responses)

> n the USA so far, copyright does not apply to LLM output in the first place.

No, this is a misunderstanding. This explains it quite well https://sfconservancy.org/blog/2026/mar/04/scotus-deny-ce...

Getting the cat back into the bag

Posted Apr 4, 2026 17:53 UTC (Sat) by vivo (subscriber, #48315) [Link] (5 responses)

Actually that article said SCOTUS refused to hear a case in which copyright was _vetoed_ for an LLM created image.
It say nothing legal (but does not negate the possibility) that code written by an LLM can be copyrighted

Getting the cat back into the bag

Posted Apr 4, 2026 18:33 UTC (Sat) by bluca (subscriber, #118303) [Link] (4 responses)

Yes? That's the point, that case is being widely misinterpreted as "In the US LLM output cannot be copyrighted", but that is a wrong interpretation, as explained in the blog post

Getting the cat back into the bag

Posted Apr 4, 2026 20:29 UTC (Sat) by anselm (subscriber, #2796) [Link] (3 responses)

The blog post explains that the output of generative AI may be copyrightable if a human being works it over enough that sufficient human creativity is involved (again). The court – like, incidentally, the US Copyright Office – refused to entertain the notion that a computer program could be the recognised sole author of a copyrighted work (which was what the guy with the picture wanted). Anthropic, however, claims that Claude Code is “100% written by Claude Code”.

Getting the cat back into the bag

Posted Apr 4, 2026 20:33 UTC (Sat) by bluca (subscriber, #118303) [Link]

No, that is your editorialization, which ironically looks like AI hallucinated output. The key of that case is that the claimant waived any claims to consider his own copyright. Did Anthropic waive all copyrights on Claude Code?

How much human involvement is needed to make it copyrightable?

Posted Apr 7, 2026 9:48 UTC (Tue) by farnz (subscriber, #17727) [Link] (1 responses)

Sufficient human creativity may not, however, be that large.

The court's judgement is that something needs to be created by a human to be eligible for copyright protection, and that because the plaintiff had explicitly disclaimed human involvement in the creation of the artifact, there was no need to consider whether the human involvement was sufficient to reach the "copyrightable" bar.

So the question for the Anthropic claim is do they mean "100% written by Claude Code" in the sense of "100% carved by Makita tools", where Claude Code is a tool that a human wields to create an artifact, or do they mean no human involvement? If they mean no human involvement, then the court judgement is relevant; if they mean that a human wields Claude Code the way a sculptor would wield a battery-operated chisel, then it's not.

How much human involvement is needed to make it copyrightable?

Posted Apr 7, 2026 11:24 UTC (Tue) by anselm (subscriber, #2796) [Link]

So the question for the Anthropic claim is do they mean "100% written by Claude Code" in the sense of "100% carved by Makita tools", where Claude Code is a tool that a human wields to create an artifact, or do they mean no human involvement? If they mean no human involvement, then the court judgement is relevant; if they mean that a human wields Claude Code the way a sculptor would wield a battery-operated chisel, then it's not.

“100% written by Claude Code” sounds much more impressive if you take it literally as “no human involvement required, Claude Code is just that good already”. It is, however, reasonable to assume that, like so much in generative-AI marketing these days, this claim is based on very charitable, if not outright peculiar, interpretations of terms like “100%”, “written”, and “by”.

Therefore Anthropic may be on the safe side as far as the copyrightability of their code is concerned, but they will still have a very hard time if they want to leverage this to get rid of all the copies of the Claude Code codebase that are out there now.