Daniel Stenberg said Anthropicâ(TM)s Mythos AI found only one low-severity vulnerability in cURL, along with a few false positives and minor bugs, despite heavy hype around the model. After reviewing the results, the cURL team concluded Mythos performed similarly to other modern AI code-analysis tools, leading Stenberg to describe the excitement around it as âoeprimarily marketingâ and âoean amazingly successful marketing stunt.â

Yes, but now you have it too complicated for 95% of the vibe coders. So they simply won't do it. Because skipping all of those steps still results in something that compiles.

They fuck up spreadsheets as well. A truckload of business-critical spreadsheets have errors in them that often go undetected for years.

It's hard enough to get actual developers to properly consider security. Not surprised at all that vibe coders don't.

Plus, of course, most of the training data is insecure to begin with.

But let them learn by fire that there's a reason actual programmers take time to ship a product, and it's not that the AI can type faster.

Problem is: We don't even know what consciousness is.

So the best we can say is if something creates the impression of having one, based on whom we attribute consciousness to, i.e. other humans. Well, big surprise that a model explicitly trained on human language and texts creates that impression. It does show just how good the models are. At pretending to be human because they have a shitload of examples on what humans would say.

For all we know, the gas clouds on Jupiter could be conscious, just in a way that is completely baffling to us. We can't rule it out because we don't know what consciousness is, so we can't test for it.

Not true.

An attacker may have a limited window. He might exploit some other vulnerability to do some operation with privileged access rights, but not have an admin shell.

You're not saying?

But they take security so seriously. They said. They promised. This time for real. No, this time. Ok, next time.

This.

Whenever a politician claims that something is "to protect the children", you can be 100%, absolutely certain that it is not about the children.

Your metaphor would make a lot more sense if kids were designing Lego sets that Lego then sold and vastly under compensated the creators of those sets (while doing a piss poor job of keeping those kids from being contacted directly by adults)

Oh, that's trivial: a) it's made-up nonsense, b) it tells people how to live their lives and c) it's been around for so long that people forgot that it's made-up nonsense.

None of that or the rest of your answer has anything to do with the point I was making: That "accepted as a religion in the USA" isn't much of an argument. If people can get Jedi accepted as a religion, it just proves how meaningless all of that is. Other countries have correctly identified Scientology as a pyramid scheme and a scam.

The fact that other religions would qualify for that as well doesn't make it any less true.

Yeah, but isn't the bar for that ridiculously low in the US? Like the Jedi "religion" being tax-exempt on religious grounds? And there's no discussion that that one is based purely on fiction.

No, I did not. These are features that exist for my convenience, not as mass surveillance tools. The government is abusing them.

Yes, if you have control of the cell towers. Or run an IMSI catcher. But the technological solution applied is not the question. The question is if we want someone as untrustworthy as our governments to be able to constantly track us.

The fact that you personally maybe don't care doesn't give you the right to opt-in all of us who might care. The problem is that once the technology exists, it will be abused. It already is. If we explicitly allow it, abuse will run rampant. We already have examples of cops using surveillance tech to spy on their spouses, or to stalk that cute girl from the bar. We have tons of examples of surveillance permissions granted for one purpose being used for another one. When the government wants these laws it's always to find child abusers and terrorists. But that is never what they actually have in mind.

I don't think it's the AI specifically, but the fact that they've used AI to let go of competent (and expensive) people.

I'm using AI as a coding assist and code reviewer myself. It is impressive how often it is spot on, but it is also impressive with how much conviction it tells you one thing, then after you correct it it admits that that was totally bonkers. AI or not, you need someone in the loop with a deep understanding of what it actually is you are trying to accomplish.

I can fully imagine an AI without guidance to go off the rails more and more over time. But I can imagine the same thing for a room full of junior programmers.

I am honestly surprised how long it took them to ruin it. I moved my projects off github years ago, expecting it to fail much sooner.

"can't tell the difference between a game and reality"

Uh, while I would argue that you should probably care because that person should be focusing on an investor meeting, it tickles me that you're suggesting somebody playing a videogame during a meeting supports the assertion that "they can't tell the difference between a game and reality".

That would probably amount to a whole lot of people who can't tell the difference between a game and reality (which I don't agree with) rather than a whole lot of people are not focusing on what they should be focusing on (which I do agree with.)