larryjw
About
- Username
- larryjw
- Joined
- Visits
- 199
- Last Active
- Roles
- member
- Points
- 3,337
- Badges
- 2
- Posts
- 1,040
Reactions
-
YouTube and Spotify also won't offer any apps on Apple Vision Pro
-
Apple Watch import ban stay opposed by ITC
-
Apple Watch import ban saga kicked off a decade ago by an early-morning email to Tim Cook
Regardless, the issue is whether there was a patent violation. How it came about is not relevant.
A patent violation is a patent violation, accidental or purposeful. Defenses can be brought, such as "there is only one way to do this" or "it's obvious from prior art", or ...
I haven't read any analysis of why there was not a licensing agreement in the first place, once litigation was a possibility. -
Ex-Apple lawyer guilty of insider trading gets a slap on the wrist, avoids prison
-
Copyright laws shouldn't apply to AI training, proposes Google
Copyright protects against copying the work of others, with the idea that one is generating a verbatim copy of a significant portion of that work. Since LLMs merely takes those words from copyrighted works and updated its database of its conditional next word probabilities, there is little chance that an authors set of words will be copied into the LLM response to a prompt.Copyright also protects derivative works. This is where it can be argued that the LLM is creating a derivative work. I think that is a stretch however.In any case I don't believe copyright is in any way a law which prevents the wholesale processing of copyrighted works for the purpose of building conditional next word probabilities databases. The harm to authors that copyright laws protect are not at all impinged by building these databases.