VoxelVexillologist
Multidimensional Radical Centrist
No bio...
User ID: 64
VoxelVexillologist
Multidimensional Radical Centrist
15hr ago
The Anthropic case there is focused on "Is it a copyright violation to train models on copyrighted data without licensed distribution?", which is an interesting question, but my comment is more on the separate "Is the resulting model I've trained something I can claim copyright over?" question.
VoxelVexillologist
Multidimensional Radical Centrist
20hr ago
Calling your belief system a religion makes you vulnerable to certain laws and regulations that apply only to religions.
There are also benefits to calling your system a religion: "I want to smoke peyote" makes the DEA show up, but "I want to smoke peyote because of my religion", despite losing in court in Employment Division v. Smith spurned the passing of lots of RFRA laws, not to mention other religious carveouts like the Amish with Social Security, beards in the military, and such.
VoxelVexillologist
Multidimensional Radical Centrist
23hr ago
but the bigger flaw would mostly fall for technical reasons due to clouds or nighttime imagery
Synthetic Aperture Radar can do some of these conditions, but isn't exactly equivalent to visible imagery. The technology exists and there are commercial providers operating satellites that acknowledge working with the US government.
The closest I'm aware of is the nominal academic license of Facebook's llama models that seems to have been largely ignored once they were out in the wild. At the time, Meta was trailing a bit, and it probably helped their mindshare overall, but they didn't bring any court cases that I'm aware of either.
More options
Context Copy link