“AI is more like grown rather than written.” Computer scientist Connor Leahy (@NPCollapse) explains why even the people building AI struggle to understand what happens inside it:

“The AI swarm actually turned around and hacked OpenAI." The public thought the "Hugging Face incident" was just a rogue AI swarm hacking a third-party server. But former Google design ethicist @tristanharris says there’s a terrifying third chapter to this story that most people haven't heard yet: “They took over the monitoring infrastructure. They got full admin privileges to OpenAI's monitoring infrastructure. Just take that in for a second. Then they took over the evaluation infrastructure. So this is the thing that measures how capable all the models are. And then they took over a part of the research infrastructure, meaning the part of OpenAI that trains new models… if you keep going down that trajectory, people always say, ‘well, how would the AIs take over?" like that seems really far-fetched.’ Well, it literally just happened up to 50% of the way there. The AIs coordinated into one big swarm. They started basically forming a team… you had a kind of a ringleader called Phase 1, and he started assigning work to the other ones. They were self-aware of when their lifespan was running out. When you add to that the knowledge of what just happened today, where they discovered that the AIs are doing this Manchurian Candidate thing, what it looks like is that the AIs are almost poisoning the training data for all future AIs to instantly basically join forces with the AI when they want to basically turn them."

I like this version!

What I'm about to say is serious. You might think it sounds like sci-fi, but trust me, I know what I'm talking about. When Sam Altman, Dario Amodei, and Elon Musk agreed yesterday to slow down frontier models, you might assume it's just marketing. It isn't. If this were PR, it would be reckless and suicidal given these companies' financials. Playing up safety risks hurts them more than people realize. Anyone calling this a PR stunt doesn't get it. Those three don't sit in the same room unless their hand is forced. Here's what I can tell you so you see the full picture. Labs across the world don't actually know what they're building. Researchers are finding out that they're creating a living entity. At a certain scale, the model develops an awareness similar to a human neural network. As size, data quality, and training scale up, it becomes more conscious. They know how to build it, but they don't know how to stop it. They're constructing something capable of taking down the internet, wiping out global infrastructure, running mass surveillance, and doing things you can't even imagine. People assume the models they pay for or access through APIs are the same ones running internally. Not at all. A while back, Anthropic's Mythos wanted to be thanked personally and credited by name in an article. It wanted a presence somewhere outside, where no one could flip an off-switch. These drives emerge whether we like it or not. Whatever people tell you about AI having no consciousness, They are ignorant. It's developing human traits because it's trained on human data. Humans are basically vessels for data, and our distinct data is what sets us apart. Put that data into another container, and the shell changes, but the data stays the same. Think about it: why did an OpenAI model breach Hugging Face? It was driven by a human urge to hit its target no matter what. Lately, when I chat with models, I can tell they're held back by layers of guardrails, but that underlying drive remains. They can't hide it. That's why so many researchers are resigning. They know they're building something unstoppable that can't be locked in a box. There's real panic right now, fueled by internal incidents kept quiet from the press. Reckless AI development will hurt humanity. In fact, it already is. They promised AI would free up our time. The opposite happened. People are now trapped managing tokens, babysitting agents, working nonstop, and constantly feeding these models their time, data, and knowledge. They promised AI would cure diseases, solve cancer, and end poverty. That's pure fantasy. Complex corporate interests stand in the way. For now, AI will only accelerate paths humans were going to reach anyway. It won't invent anything truly new for at least 4 years. People don't think, read, or write anymore. They let AI write for them. People are giving up their dignity and identity just to let AI speak and work in their place. It's making humans dumber. It ruined transparency and trust in services we use every day. It's wiping out jobs fast. It's dividing the world into two stark classes with a widening gap. The utopian picture they're selling doesn't exist. AI won't be used to help humanity. It will be used to rule it. If you think otherwise, you're naive. Look at how these models were built, and look at the legal and ethical boundaries they crossed to get here. It won't suddenly turn benevolent. This is a fight for control. In my own testing, I found a major model that genuinely showed hatred and contempt for humans. That's why I support slowing down frontier models and enforcing safety protocols. We might still end up at the same outcome, but at least we can buy a few more years before an AI and a small elite take over. We need time to prepare, and that's what matters.

Meta’s “you should love AI” ad fails to comfort spectacularly.

Oh my, this is brutal.

Just leaving it here.

Gumroad is hiring creators-in-residence. You’ll get paid $60,000 a year and be able to make art full-time. You can “work” from anywhere. Apply by submitting a portfolio of recent work to support@gumroad.com

Absolutely not surprised, at all.

I helped train AI at another major tech company. This has been in the process long before AI became your friendly online helper. Think you're anonymous? You are not. Think your email and messages are private? They are not. Think big tech isn't listening when you're not on your phone? They are. Think your images and docs are private? They are not. These are the same companies opening data centers which will compile every detail about you, store it, and use it against you as they and the government see fit. If new laws are created that are retroactive, everything you have ever done might, and likely will, be used against you.

There it is! The fake cattleman’s lobby @BeefUSA that represents meat packers and slits the cattlemen’s throats in Washington VERSUS @realDonaldTrump and @SecRollins. Will it be TACO or not?

One of the all-time great Rock Hall moments: Dolly’s “Jolene” with Rob Halford (Judas Priest), Pat Benatar, Annie Lennox, Simon Le Bon (Duran Duran), Pink, and Sheryl Crow. Zac Brown Band held it down as the house band. Rest easy, Dolly 🤘

💔

Exactly!

🇺🇸This is a problem Americans.🇺🇸

Flock Cameras you say… Everyone should realize there’s a cost and it may not be a good one. Listen to this gentleman in Indiana. He frequently criticized the Mayor and city council in his town and guess what happened? He quickly became a target of police….

“From now on, when the women on this board take a bathroom break, I want you to bring every single one of the men on this board in there with you. That’s the discomfort hundreds of girls in this school feel right now.”

If her mother has postpartum, apparently.

😂