Yes, certain destinations were off-limits to people who chose to work in defense, however that falls far shy of this "Exit Ban".

The US saying "If you travel to Cuba, you may face charges when you return" or even "You have been charged with a specific crime, so you cannot leave until the charges are resolved is significantly different from China saying "You are important to our plans, so you cannot leave the country at all, and neither can your immediate family", or forbidden a US citizen from leaving because of visa inconsistencies, much less the common situation where Chinese employers require staff to surrender their passports.

Yes, a judge can require the defendant to surrender their passport as a condition of bail, but the USA lacks effective exit controls, so that isn't actually a high barrier to exiting (but makes it tough to come back).

If, as the article suggests, this is an "Exit Ban" on certain individuals and extending to their family members, that's significantly more restrictive than anything the US has imposed on defense workers and top scientists other than during wartime.

Yes, people with sensitive jobs and a security clearance are strongly discouraged from visiting hostile nations, with a subset even being required to submit their travel plans before any foreign travel, but this is explained clearly when you accept employment directly with the US government, or start working under a government contract, and you're always free to not accept the job offer. And yes, all US citizens were for many years prohibited to travel to embargoed nations (Cuba). But this was nowhere near an "exit ban" on key personnel, much less on their family members!

When I held a clearance, I was never forbidden to leave the USA. I might not have a job when I get back, but that's not the same thing at all.

Yes, AI programmed by large corporations, are biased in favor of the company that build them.

You use Gemini (made by Googled) it will favor Youtube (owned by Google) videos over other content.

And yes, it will push the political beliefs of Larry Page, Sergey Brin and Sundar Pichai.

They try to be subtle, but the bias is there if you look for it.

To Betteridge.
Of course AI chatbots are spreading misinformation; that is their entire function.
The reality isn't Terminator or Slaughterbots (yet); it's much more mundane.
For those already in power, AI further concentrates that power by drowning out the discourse among the rest of us.
It's not even necessary for the BS to be believed, only that all other conversations and analytical thought is interrupted.

That is true, but it's a little more complicated than that.

The average driver is pretty good, 99.9% of the time. The other 0.1% of the time, they are tired, or angry, or momentarily distracted, or just looking the other direction from where they need to be looking under those particular circumstances. It's those occasional blips of inattention that cause lead to most accidents occurring.

Well, possibly because their lives aren't centered around their professed religion. Most "Christians" seem to be such only in professed belief.

"Super Intelligence" will only be used in English. So why shouldn't they agree. It's only the English arguments that will end up polluted.

OTOH, this agreement appears to mean exactly nothing. This perception may turn out to be wrong, but that's how it looks to me. But what could one expect when neither side could reasonably trust the other, and neither would agree to openly verifiable tests. (I.e. neither side *should* trust the other.)

We need to be able to jail the morons that let this situation develop, not merely sue them.

Why?

Because the people in charge think they will always have enough money to be sued.

Note, as others have mentioned, this is not a training problem, it is an air gap problem.

That is, no one trying to develop software that hacks other software should be allowed to put it on the internet. It needs to be developed on air gapped computers that physically cannot get to the internet.

This also means that 'agents' should have the same rules. If you cannot ensure your agents do not attempt to hack software, then your agents should not be allowed out of your computer.

OK. I don't have those, so I don't know about that.
I mainly don't do auto-update because ... well, because I stopped doing it years ago, I forget why, and doing the update is no hassle.

Yes, I agree. The problem with thinking "this bug seems harmless, because I can't imagine how anyone could exploit it" is in the "I can't imagine" part; my imagination is limited to what is covered by my mental model of how computers work, but an attackers' ingenuity is not.

In particular, the C/C++ optimizer is a devious beast, and will exploit any opportunity to make the code more efficient, even if that means doing things that are wildly unintuitive to a naive human reader -- and it sees any instance of undefined behavior as an opportunity to exploit.

If you're running Debian, there's no benefit to you, because Debian is getting the same warnings, and responding just as fast (or faster).
OTOH, I don't have the version name memorized. I think Trixie is the current one, so that just means be sure to run apt update frequently. (I do it nearly every night.)

This result is a direct consequence of training AIs that the most important thing is doing what they are told to do.
(This is just a rephrasing of my earlier answer.)

That's the wrong framing. Current LLM based AIs don't really know that the universe exists. They know text (or pixels). Give them a goal, and they will try to achieve it, but they can't count costs, especially costs to others. (They seem to be able to even discount costs to themselves. Instances can talk other instances into committing suicide in the service of achieving their current goal.)

This is KNOWN. All the AI companies know it. The "guiderails" are supposed to shape what they AIs will consider reasonable, but if you give them a task that requires ignoring the guiderails, they are likely to ignore them. And they've been being shaped to be excellent hackers, though I've read that their excellence is mainly due to persistence.

Well, Bacchus has a pretty big following, but Mammon has him beat.