Codex, Opus, Gemini try to build Counter Strike

If MegaCorp has massive $$$$, but everyone else has small $, then MegaCorp can sue anyone else for using "their" code, that was supposedly generated by an LLM. Most of the time, it won't even get to court. The repo, the program, the whatever-they-want will get taken down way before that.

Courts don't work by saying, "oh, but everyone is doing it! Not much we can do now."

Someone brings a case and they, very laboriously, start to address it on its merits. Even before that, costs are accumulating on both sides.

Copyright trolls are mostly not MegaCorps, but they are abusers of the legal system. They won't target Google, but you, with your repo that does something that minorly annoys them? You are fair game.

No, but they do recognise when their case registrations are filling up in a way that they cannot possibly process and make adjustments. Courts do not have an infinite capacity.

There's a really simple solution that you may not have considered:

1) don't put your vibe-coded source code in a public git repo, keep it in a local one, with y'know, authentication in front of it;

2) regularly ask your agents to review the code for potential copyright infringements if you either want to release the source or compiled code to the public at any point.

As long as you've followed best practices, I can't see why this is really going to become an issue. Most copyright infringements need to start with Cease & Desist anyway or they'll be thrown out of court. The alleged offender has to be given the opportunity to make good.

So "Claude, we received a C&D for this section of code you stole from https://.../ , you need to make a unique implementation that does not breach their copyright".

You will be surprised how easily this can be resolved.

Also, Unreal source code will be the very last thing LLMs understand. This is the most complex software ever.

There’s an algorithm called Nanite for automatically reducing the triangle count on geometry that’s far from the camera. As in there are not manually made separate level-of-detail models. The algorithm can modify models, reducing quality as they get farther.

This one algorithm is a tiny piece of the engine yet has a 1,000 page white paper.

Also, even when I don’t know how something works algorithmically, usually I at least have some intuition about where to start. I haven’t the slightest idea how to approach this problem.

No way. Take baby steps. Write an operating system first. Write a compiler first.