You can run openclaw locally against ollama if you want. But the models that are distilled/quantized enough to run on consumer hardware can have considerably poorer quality than full models.

Also more vulnerable to prompt injection than the frontier models, which are still vulnerable, but less so.

This is actually the scariest part of the article for me.

It's clear we've got to the point where at a glance it is hard for those who are otherwise unaware to tell the difference between AI slop and organic content.

If nerds on HN can't tell the difference between an AI slop influencer and a fairly well-regarded human influencer... how can we expect the rest of the public to tell the difference when it comes to science, health, civics, politics, etc???

We're at the cusp of a distrust and misinformation cliff that is going to be terrifying in magnitude.

The article didn't suggest that the video mentioned was AI slop, it correctly recognised it as human generated.

I know he said it was not AI, but he but still described it as “slop”, lumping it in with the other examples. And said it was a video “where a woman decides to intentionally start a fight with her boyfriend” which isn’t really an accurate description. She’s a well known comedian playing an obviously exaggerated character that pokes fun at relationship dynamics.

My point here isn’t simply that “people can’t differentiate between AI and not AI” (although that is an issue for some) but that the prevalence of AI slop lowers the trust of ALL content even when they know it isn’t AI generated. This author was so fed up with the content they were being served that they were quick to dismiss other content along with it at a cursory glance.

Indeed. He thought it was not AI slop, but the kind of low-effort slop ruining Facebook.

Your opinions may vary, but this is not one of those super clickbaity social media personalities; people like her because she's funny.

I suspect that the direction of these situations often depends on how your initial email is routed internally in these organizations. If they go to a lawyer first, you will get someone who tries to fix things with the application of the law. If it goes to an engineer first, you will get someone who tries to fix it with an application of engineering. If it were me, I would have avoided involving third party regulators in the initial contact at least.

Yes, this routing is common. German energy company recommended by a climate organization had a somewhat similar vulnerability and no security contact, so I call them up and.. mhm, yes, okay, is that l-e-g-a-l-@-company-dot-de? You don't want me to just send it to the IT department that can fix it? Okay I see, they will put it through, yes, thank you, bye for now!

Was a bit of a "oh god what am I getting into again" moment (also considering I don't speak legal-level German), but I knew they had nothing to stand on if they did file a complaint or court case so I followed through and they just thanked me for the report in the end and fixed it reasonably promptly. No stickers or maybe a discount as a customer, but oh well, no lawsuit either :)

In the early internet days, you could email root@company.com about a website bug, and somebody might reply.

> If it were me, I would have avoided involving third party regulators in the initial contact at least.

I'm surprised to see this take only mentioned once in this thread. I think people here are not aware of the sheer amount of fraud in the "bug bounty" space. As soon as you have a public product you get at least 1 of these attempts per week of someone trying to shake you down for a disclosure that they'll disclose after you pay them something. Typically you just report them as spam and move on.

But if I got one that had some credible evidence of them reporting me to a government agency already, I'd immediately get a lawyer to send a cease and desist.

It seems like OP was trying to be a by the book law abiding citizen, but the sheer amount of fraud in this space makes it really hard to tell the difference from a cold email.

Bluesky?

That's a butterfly.

In a corporate environment no-camera/no-phone policies are sometimes also used for DLP reasons, out of expediency. Oftentimes it is more profitable to hire less trustworthy people (read: cheap labor) and simply make it inconvenient to steal data. This usually works good enough when you're trying to protect widget designs and not human lives.

Yeah, this is why any high security information facility has physical security controls. Give someone infinite time and physical access and they could copy it off with clay tablets and chisels.

Then you fix that loophole by subtlety altering the phrasing or formatting that you send everyone

That's why I said you paraphrase, rather than using the exact phrasing and formatting of the original doc.

Include slightly different details in each version. Then if the paraphrase mentions one of them, you've identified the source.

Yes, I'm aware of that approach.

It's likely tougher than it seems; the big important bits that the news will care about have to match up when checked, and anyone with high-level access to this stuff likely has a significantly sized staff who also has access to it. Paraphrasing reduces the chance of some minute detail tweak being included in the reporting at all.

You also have to actively expect and plan to do it in advance, which takes a lot of labor, time, and chances of people comparing notes and saying "what the fuck, we're being tested". You can't canary trap after the leak.

The model files and designs are protected speech, yes.... but free speech doesn't really follow the transitive property. Just because you can have a copy of the Anarchists Cookbook doesn't mean that building the things in the book are protected speech.

The state applying strict scrutiny is an intimidation tactic that infringes legitimate speech. We also know very well that none of these systems will work well and that the DOJ will be happy to apply bans as far as they can get away with. It will be the no fly list for physical objects, and we won't even be told what is explicitly banned. We'll just not be able to print something. It will be very hard to argue to argue to the courts and at that point the harm is done, and there is no real effective remedy. They've already blocked and discouraged a legitimate right.

There are thousands of physical objects for which possession is already controlled.

Which physical objects that you make in your home, are actively monitored by the police as you make them? When I bake something in my oven, would it be reasonable for the DEA to have a camera in my kitchen to ensure I do not make weed brownies?

I think you are a little confused because the title is a little bit editorialized, but, this law doesn’t monitor anything in anyone’s home. It controls the commercial sales of 3d printers.

No, I am not.

"This bill would require, on or before July 1, 2028, any business that produces or manufactures 3-dimensional printers for sale or transfer in California to submit to the department an attestation for each make and model of printer they intend to make available for sale or transfer in California, confirming, among other things, that the manufacturer has equipped that make and model with a certified firearm blueprint detection algorithm."

This in my opinion is the state monitoring me in my own home. You can split hairs on semantics, but the effect is the same if machine learning or a person does it.

Not in California but generally 3d printing or otherwise self-manufacturing firearms is legal per the 2nd amendment, and publishing files is protected expressive activity (I bet there are some ITAR cutouts for weapons but I suspect that doesn't come into play here).

Self-manufacturing is definitely federally legal by statute but I don't think it is entirely clear that a constitutional right to self-manufacturing is incorporated to states or localities. Heller never mentioned manufacturing.

And on top of that, there is no protected right to 3d printers.

Regardless, it is 1000% more clear that states can't ban you from sending around gun design files.

Bruen requires a historical analogue: home manufacture of arms is practically an American tradition.

There are three types of people:

1. people who don't have a gun

2. "gun owners"... who have them as a hobby and have a bunch of guns

3. people who just happen to have a gun at home. you don't hear about these people because they aren't "into" guns and don't talk about them.

I know people who don't like guns don't want to hear it, but guns are tools. You wouldn't ask a carpenter if the only tool he needed was one saw. You don't just need a saw, you need a hammer, a screwdriver, a drill. And you don't just need one saw, you need a few specialized saws, and many screwdrivers. Maybe one hammer is enough.

Yes, and while most people aren't carpenters, some people are interested in carpentry, and some people just need to cut one board.

I am #3 and there are a lot of us (every close friend I have is #3). we don't talk about them publicly so every non-close friend I have would lose a house betting against me owning a gun (I am as non-conservative politically as it gets)

Liberals and guns are like conservatives and abortions. They both have them they just don't talk about them.

Going after model distribution is by far the least practical option. Speech is protected by law that California can not change, and the flow of information is ridiculously difficult to control even if it were legally permissible to do.