The target audience for this particular rule of the AI Act isn't the producers or users of AI. It is the consumers of content that might be AI generated.

And frankly, letting people know if they see AI content or not is a fair thing to ask.

Fun fact: The EU AI Act considers this. It makes a difference between "AI generated" and "AI modified".

But it's hard to put this into the type of watermarking they are using. Because if you generate a text and then change a few things here and there what you are doing is degrading the watermark. But since the watermark is probabilistic, it's hard to tell if it was a weak watermark from the start or if it was degraded.

That ain't the same thing.

This thing is doing things you didn't tell it to. It might even do things you explicitly told it to not do. It is not a case of "well, you could've forseen this as a consequence if you had been careful" like your campfire. My example is only a bit silly, it is absolutely possible that an AI agent will decide to hack NASA because you asked it for some data for a school project. Even in the broadest possible sense, that does not fall into stuff that the law so far has considered to be in the domain of negliegence.

AI agents do stuff you could not possibly have expected. Heck, I asked Claude a fairly simple knowledge question yesterday and it wanted to read from /dev/null. There is no argument you can make that would convince a jury that I should in any way have expected that and taken precautions. In a hypothetical world where a read from /dev/null randomly kills a puppy somewhere on the planet, I don't think I would've been convicted.

it also presumes YOUR device is the most important device around, as opposed to, say, air traffic and flight scheduling systems, train systems, electricity control systems, stoplight control systems (they adjust their 'rush hour' times each switch), cruise and shipping departure and arrival times.

All of those systems need independent testing time to confirm that they have the switch, they can exchange information with other computers and be accurate about the switch, and all handle the fact that there would be no 'extra hour' on Nov 1st that would have already been accounted for in their schedules for that day...and note they'd be dealing with computers in other countries that would likely be even slower in getting the upgrades needed to test on their end.

But that's not what Trump is asking for. He is asking for "effective immediately", just like the British Columbia bill was.

Yeah, you're suggesting Congress could set it for next year (giving next year's Congress a chance to revoke it), but that's not what is being proposed by the Administration nor the bill that already passed the House but the Senate seems to be never picking up.

that is a very naive way of looking at how computers work, especially computers that have to share future scheduling information with other computers. NO computer has what you are talking about that would just magically solve everything - especially in THIS case where the DST would be 365. Yeah, you could tell your computer to 'lie' about its location and say it was in, say, Brazil rather than Chicago in order to stick to whatever TZ you want your device to think it is in...but that is not something that scales up to all of the hundreds of data centers on the east coast with the millions of computers involved. They need to be absolutely and automatically accurate to where they are and 'when' they are, and no 'flip of a switch' will be adequate for that purpose.

Hell, about half the computers and software systems (like databases) in the public right now still aren't ready to hold Americas/Vancouver time on DST-365 (my own macbook didn't get the upgrade until about a month ago) when the flip happens. Not all the upgrades have rolled out or people are still intentionally using older systems because they "just work" (e.g., sticking with Java 8 and not having any time or inclination to try to get to Java 25 which changes so many network related libraries one might as well just rebuild the enter system from scratch).

A calendar-clock change like this needs at least a year of time to prepare it and get all the software systems tested and ready for the change when it happens. Yeah, it is easier to actually code the changes now for some systems than it was when we had the big change in 2005-2006 (in fact, because of dealing with that change), but that doesn't simplify the testing required for something on this scale.

You don't get it.

Autonomous AI agents are the perfect no-responsibility card. The AI companies? They trained a general-purpose model and never gave it any explicit instructions to do hacking. The AI services? They are just middlemen and have no control over what people tell the agents to do. The people prompting the agents? Unless you're straighforward dumb, you didn't tell it to do hacking, you just told it something innocent like "find me all data on Mars for my school project" and the agent decided that probably NASA is hiding something so let's hack them to get it.

Since there's no human anywhere executing anything that's a crime - voila, in addition to the victimless crime we now have the perpetratorless crime.

https://threadreaderapp.com/th... yeah - the mass-buying started back in June, months before any announce or post was ever made on the matter. And yes, several are now selling for millions.

per a twitter post, someone did some searching and found that .si (Slovenia, yes where Melania's from) domains were being bought at an average of about 400 / month, usually with words in their own language.

But starting in *JUNE* (months before this announcement or any other like it), scalpers started buying up .si domains with common and corporate english words by the thousands...and now those domains are selling for millions.

Somebody knew it was coming, months ago. This was very well planned out to make some people very rich, possibly including Trump O itself, or those who buy his NFTs/Coins. That whole paying to get private access? This is what they're told to do in those back-door meetings.

https://threadreaderapp.com/th...

not for me. i still see AI in that first large text content: "America.gov

  uses AI
to get simple answers only
from

  official
government sources."

A resounding success. No large-scale demonstrations, damaged cameras, lawsuits or political suicide for those involved. So we can expect this to be rolled out fully.

Crime prevention? Who said this has anything to do with crime?

Yes, limited local AI models certainly are going to be a standard feature in the future.

I run both local and cloud AI. There is still a huge difference, because my local machine has to use smaller and quantized models. It takes considerably longer for lower quality results.

because i note - you aren't hearing this from Bezos, Ellison, nor Alphabet.

another - "we have enough AI, we shouldn't have more" - a way for the big 3 to try to get Congress or local governments to act...against the other vendors still building out more data centers and developing their own tech (Oracle, Amazon, Google).

"let us keep the monopoly on it all and leave those others behind, and /maybe/ we can control it...but if you let too much competition in, then our fears will happen."