This page claims over 400,000 recordings but links to a listing of only 187,034 audio files. I'm guessing the discrepancy is the girth of the suit: IA agreed to take down the files that the plaintiffs could prove were theirs and no money changed hands.

There's your problem, right there.

There is no such thing as complete delegation. The word for that is abdication.

Let's compare, shall we?

Little Bobby Tables:

• No framework required: conventional database entry + payload only
• Wreaks havoc in an instant
• Total size: 32 bytes

This:

• Downloads ollama (672 MB, on Windows)
• Downloads a 14 GB data file for the model itself
• Requires a bare minimum of 16 GB of VRAM—and still runs like absolute molasses, eating up all resources
• Total size: 15 GB

Personally, I'm on Team Tables here. Maybe in a decade or three this will be practical.

Slaughterbots

OK. There are malicious crawlers out there (for AI or other things) that ignore robots.txt, but the big four don't ignore it.

If your website is being murdered by crawlers, stop them.

Too obvious? What am I missing?

I heard FAT12 had it beat for raw performance.

"Hello, Ecuadorian embassy? Yes, I heard about the weird shit with the cat, but listen..."

Also possible. OpenAI's models do love their bullet points, and especially having exactly three examples in a list. The catchphrase "math magic" at the end also reeks of GPT-4's didactic blandishment.

Yeah, there aren't a lot of popular map projections that fail that test, unless you perhaps count the polar projection used for the UN logo. I'm guessing that sentence is there as a result of editing; perhaps it originally said equally-spaced parallels (which is true of the Robinson projection) but someone math-savvy was consulted to correct the claim to its current form (without seeing the context) and the maintainer of the page wasn't knowledgeable enough to realise it should just be removed.

For some of it, yes. We compete in the SaaS space, and we switch customers back and forth with our competitors all the time. We drop 'unique data' into their databases when we're custodians so we have a chance to spot misconduct if/when the customer switches to a different SaaS provider.
We see a fair number of them surfacing in the wild, and quite often in major brand providers. These are either undeclared breaches, or something worse.
We're not about to start blowing whistles on specific cases. We're not really geared up to go to war with companies that have billion dollar legal departments.
Suffice to say, anyone who thinks their data is safe in the Cloud is optimistic at best. Their data _should_ be safe, but it is far from it.

You started out well, but you're condescending in your first point. Not likely to get a good reaction from anyone there.
Then, insulting in your second point. You're starting to look like a troll here.
More insults in your third point. Getting a real troll vibe from you now.
And then we get into the ad hominem. Definite troll.
Shame really. You might have had some good points, but your tone trashed it. Next time, get ChatGPT to sanitise your responses before posting.

The thing is... there have been plenty of outspoken and highly visible survivors of school shootings for decades. How the hell do we exorcise this gray paste of untruth?

In English, "IA" is usually understood to mean "Internet Archive." Only "AI" means "Artificial Intelligence."

Its easy to answer when you realise that the Cloud is just someone else's computer. When you stop paying them, they'll do whatever they like with your data.

In Oz, there are privacy laws that are intended to protect the data, but most of our competitors openly ignore them.

In the EU, the GDPR is intended to protect the data, but once again, no one is policing it.

I don't know what the situation is in the US, but given the conduct of the 4 majors in AI development, I expect that its pretty much nothing.

Possession is nine tenths of the law, and once you give up possession of your data to someone else, you get what you get.