Actually, it has to be 'substantial human input' to be considered valid. Reworking prompts, and then doing photoshop tweaking to the resulting image is not enough to clear the bar to get copyright protection for example. Another that's been through the process, using AI to generate a book and then doing some light editing on it, the 'author' was allowed to claim copyright on the layout, but not the content after appealing multiple times and trying to claim an ADA exception. So if AI spat out 100 lines of code, and you touched up a few... unlikely to be considered enough to qualify for copyright. If you modify it enough to clear that bar... what did the AI really do?

So, in the US machine generated content isn't eligible for copyright. What does that do to a project under GPL or other licenses? Does the license become unenforceable once a certain threshold of non-copyrightable content is in the project?

Those words don't go together as a sentence like you think they do. Now, if you want to talk about a mach microkernel with a userland originally derived from FreeBSD you'd make more sense. No linux involved.

So... in the spirit of openness they'll release the training material list, right?

Free, as long as you have an internet connection to receive certificate updates to allow you to decrypt it. That's the bit that's driving this.

If I need an AI assistant to "download and launch games" then you need to fix your UI design Microsoft.

Most people that have solar on the roof are just setup for grid tie, so when the power goes out, their system shuts down as well, so they're just as in the dark as those around them. Solar + battery that can run independent of the grid is far rarer due to the increased install cost and concern over batter replacement costs down the line.

VDO.Ninja is a suitable backend for this, video goes point to point, the backend just does the initial signaling between endpoints. You can also self-host the backend if you want. Works pretty much anywhere a browser has permission to see a camera.

Or require DNSSEC, already supported by the OS?

And then access to that repository becomes a gatekeeper for offering anything for those platforms.

You can disable it, today. As per the EU directive, you won't be able to disable it on new vehicles going forward. There are a lot of safety features available, I use many, but I also prefer to have control over them which I currently do.

Yeah, see, reading is important. The change is going forward, you can't disable it, and there IS an option for systems that actively slow you back down to the current limit.

Google, a company based on their search technology first has never had a good search engine in their mail or docs clients. Well, now I know why, they were waiting to have a data slurping means of powering it instead of just making the existing implementation decent. They've already got the data, no reason to suck it into yet another mess, bah, something else I'll need to figure out how to disable.

Look at existing instances that have already been through the system: Artist worked at a prompt for awhile to create a base scene with Midjourney, then went in after and did some moderate adjusting with traditional tools ala Photoshop, etc. Copyright office said no go. Look up "Theatre D'opera Spatial" for the details. The graphic novel "Zarya of the Dawn" has also had the copyright denied for the images in the graphic novel because again, machine generated.

In the US the Copyright office has been VERY clear on this, machine generated content cannot receive copyright protections. AI prompts cannot receive copyright protection. Content that has mixed human and machine generated content must have the machine generated portions declared so copyright protection can be excluded from those portions.

The US Patent office is on the same page, patents must be human thought up and authored, full stop.