Post

Many of my prompts start with me feeding in the entirety of my existing code - hard to see how that decision relating to image generation applies to LLMs for code

Is this limited to "one-shot" workflows? What if, i.e., I correct some decisions while letting some of it rip? (I suspect the answer is "it depends, evidentiary hearings would be needed")

It is an interesting question as to how much human involvement is required. There was a patent cases as well (older, pre-agent). I would imaging the browser or Next JS reimplementations (with no human involvement) are likely pretty close to the image copyright case.

@lawfaremedia.org is my first stop to look for legal discussions. Perhaps they can pick up this week's story Link to "copyright" search on their website: www.lawfaremedia.org/search-resul...

There's a pretty high bar for copyright over the output, at least in the US: www.copyright.gov/ai/Copyright... Explicitly says that prompts and describing what you desire (i.e specs) aren't considered expressive enough. You have to exhibit substantial control over the outputs