Yeah, the big bad AI made you plagiarize a whole project down to the name, and lie to everyone about the review process. Not buying any of it.

> down to the name

And an identical bug, per the Bluesky convo Gruber linked.

https://bsky.app/profile/mmmeh.bsky.social/post/3mslc3b3u4c2...

Why not?

There have been quite a few copyright cases alleging this behavior that LLMs are not substantially transformative[1]. Note that cases have largely been ruled in the favor of the AI companies, with courts ruling that model output is “substantially transformative” of training material (although what is currently legally permissible may not be what we feel is socially permissible or ethical). The the fact that these allegations continue to be made, and the fact that AI companies seem to be trying to cover their asses by disallowing prompts which would be likely to create content highly similar to that of copyrighted training material [2] might be enough to at least warrant some consideration that Godier is being honest here.

Not that this would absolve Godier (quite the opposite in my mind- low effort ai workflows produce slop, and he chose this intentionally), but perhaps some of the onerousness here also falls on the AI companies pillaging of work.

[1] https://copyrightalliance.org/ai-copyright-lawsuit-developme... E.g. https://nytco-assets.nytimes.com/2023/12/NYT_Complaint_Dec20...

[2] https://arstechnica.com/ai/2026/07/chatgpt-stops-cloning-fam...

And it’s even more stupid because he just could have asked Claude to build it from scratch. I mean, it’s not Rocket science and all of the resources needed are in the public domain. It would be done in 3 days. 4 hours if you want to ship hot garbage.

"The dog plagiarized my homework"