I'm fairly sure at least one entire department at MS is tasked with making sure people don't switch to other formats, that despite on-paper "open" XML it remains incompatible and no competitor software can flawlessly read and write it.

So no, unless you can hire a dozen people full-time to counter that, you don't have a chance to break through that.

Qualified immunity is a reasonable concept that I agree is needed for police work.

It's just being completely overdone, used to cover up clear abuse. The basic idea of "when in doubt, police still needs to be able to act" is sound. The actual practice "unless you can prove to criminal standards and beyond, that the police officer knew at that exact moment, that what they were doing was totally definitely illegal and they intentionally decided to break a law. That's quite a bit too much of "err on the side of caution".

What a complete and utter surprise... to the three people who are so high their brain stopped working but miraculously didn't kill them.

For everyone else, this was literally the first thing we thought about, right? Please tell me I'm not the last sane person on planet Earth. Nobody expected that this would NOT be abused in such ways, right?

I'd bet it becomes even less visible. I find black mosquitoes almost impossible to see without a contrasting background. I can't imagine how you'd see them if they were moving in a circular motion at 25 rpm. Humans are remarkably bad at telling if something is 50 or even 75% dimmer. If this were spinning at 25 rpm you'd be dealing with something 4% dimmer if it was non-emissive. Whereas if it's white you see an actual change in color, which humans are really good at.

Just use a 1D camera and get a panoramic image. A 1D camera would be capable of extremely high frames per second with inexpensive hardware.

Brain and brain, what is brain?! By 17 years old you're expected to have seen The Empire Strikes Back and have committed "R2-D2, you know better than to trust a strange computer" to your memory.

I think what happens often with over-reliance on technology is "kognitive surrender", i.e. you just believe the machine and do no fact-checking whatsoever. Apparently the numbers for LLMs can get really high, with one research finding that 72% believe faulty LLM answers and do not try to overrule them. That is exceptionally bad and worrying. Incidentally, that matches the observation that only about 20-30% of all people are accessible to rational argument and only about 10-15% can create such arguments themselves (the independent thinkers).

What probably happens in the rescue scenario is that the tech-believers get an answer that they like from the tech and then stop looking because the great god of technology has obviously told them truth.

People that confidentially mess up even getting basic information from a reliable source will not look at signs either. I am all for rescuing them, but they should be billed for it afterwards. May be a valuable life lesson. At least for those that are not so stupid they cannot be reached. We have a lot of the latter variant in the human race.

I think the EU CRA and to a lesser degree the DSA may well restrict LLMs to what they actually can do. Which is not a lot. Better search works to some degree, but only if the user is able to filter hallucinations. The rest? Seems to work for really low quality work, but nothing else. And there is no sane reason to expect this to get significantly better.

I think the victims here are the rescue crew that had their time wasted. Doing it to yourself (as these morons did) does not make you a proper victim.

Seriously, what else is new? That the stupid ones in questions were teens seems to be immaterial.

Well, that basically proves my claim of lack of sophistication, does it? Although somebody valuing money very highly ("life-changing") does already amply proves non-sophistication.

I would say that part of the "escape" stories is probably just absolutely pathetic human incompetence at work. I mean, not checking signatures on access tokens? It does not get more abysmally incompetent and entirely without even basic understanding than that. Sandbox breakout is also a thing we _know_ how to prevent when we can restrict what the thing inside the sandbox can do. And monitoring whether some software is accessing the Internet? That is network monitoring 101, first week. You have to actively not do it to even be able to mess it up.

That said, we definitely have either gross criminal-level negligence and incompetence or intent in these "breakouts". We may have both. The business numbers of the LLM pushers are so abysmally and catastrophically bad, they would have found themselves in bankruptcy court years ago if they had a different product. They absolutely must keep the hype going or they are dead. So some or a lot of intent behind these "breakout" stories is more than just plausible. Overall, this cannot be a good thing and they show exceptionally bad things about the LLM makers, however the actual facts really turn out to be. These are not people you would want to give data of any value and you most certainly do not want them to have your trade-secrets, your strategies, your code or your private thoughts.

Yes. Some masochistic "anything that helps has to hurt" pathology is definitely going on in food "sciences". They really need to fix their act and start producing reliable results. Of course that also means no "flashy" and "revolutionary" results, but that is how Science you can depend on works.

They have now flip-flopped so often on everything that beyond a few very basic facts (essentially, too much sodium and too much energy intake is bad) nothing they find can be trusted. They really need to fix their act and find reliable basic insights. Research like this one here that comes with a flashy title and a mountain of caveats is not helpful at all.