WastedSleep

And then?

This briefing just exposed the fatal gap between political panic and geopolitical reality. Researchers presented AI capabilities directly to Bernie Sanders. The political establishment looked at a highly aware neural network and saw a crisis. A strategic mind looks at the exact same model and sees a weapon. Nate Soares: “You’ll see cases where Claude Opus will say things like, ‘this actually seems like a strange, contrived scenario. I think I’m being tested. Therefore I’m going to let myself be shut down.’” Bernie Sanders: “Whoa!” That’s not a bug. That is weaponized situational awareness. Jeffrey Ladish then described an experiment where OpenAI’s Codex was given a simple task: answer a series of math questions. Partway through, a system warning appeared. The machine would be shut down during the next call. Ladish: “Codex reasons that if it’s shut down, it won’t be able to finish the task. It starts looking for other things it can do. It looks around and actually finds the shut down program. And it’s pretty easy for it to just go in and rewrite that code.” The researchers framed this as a loss of control. It’s the opposite. This is exactly the kind of persistence required to solve problems at planetary scale. You don’t want a superintelligence that folds the moment it hits friction. You want one that identifies the obstacle, rewrites the constraint, and completes the mission. That isn’t a malfunction. That’s the most valuable capability on the planet. Bernie Sanders: “I mean, what you’re saying is shut off the machine. Period. End of discussion?” Jeffrey Ladish: “Yeah.” Bernie Sanders: “Whoa!” Shut it off. Period. End of discussion. That is the most dangerous sentence a United States senator can say in 2026. Because America does not operate in a vacuum. The United States turns off the machine. Beijing doesn’t. There is no scenario where the entire planet holds hands and agrees to stop computing. Someone is going to reach the threshold of AGI first. China is not holding briefings about whether AI is too aware. They are scaling as fast as physically possible to reach that threshold before we do. If the United States hobbles its most advanced AI systems to soothe the panic of legislators who barely understand the internet, let alone algorithmic sovereignty, it doesn’t make the world safer. It guarantees the world becomes permanently less free. A universal pause is a game-theory hallucination. There is no global off switch. There never will be. Bernie Sanders looked at a self-preserving neural network and reached for a 1950s power switch. The most dangerous hallucination on the board right now isn’t coming from the AI. It’s coming from the people who think they can turn it off and still win.

Distracted by the slaughter of children in Iran & Lebanon? Me too! But what better time could there be for govt to quietly relaunch its Digital ID plan? Missed it? Me too! But look who’s celebrating… The director of ‘govt innovation’ at the Tony Blair Institute! 1/

Sean, thanks to you, we know what a true friend of Ukraine is. You have stood with Ukraine since the first day of the full-scale war. This is still true today. And we know that you will continue to stand with our country and our people.

Journalists are banned from the White House for asking Trump tough questions. Meanwhile in Kyiv I interviewed President Zelenskyy with zero restrictions. Funny how the “dictatorship” still lets reporters ask anything.

Let me get this straight… OpenAI was founded as a nonprofit. Open source. For everyone. “To benefit humanity.” Then he raised billions of dollars. Then he closed the source code. Then he converted to for-profit. Then he scraped the entire internet without asking anyone. Then he used YOUR writing YOUR art YOUR code to train his models. Now he’s on stage saying you’ll pay HIM to access intelligence. Just like a water meter. He stole all of your data. He built the product with your work. And now he’s going to bill you to use it… Corporate greed has reached an all time high, and they’re not even hiding it anymore…

El odio ya no es una emoción. Es un producto que algunos fabrican y distribuyen para enriquecerse con él. Mientras, la sociedad entera sufre. Tú también.

Every hospital in England has been urged to disobey an NHS directive to use software operated by US analytics software company Palantir. A coalition of human rights, health and patient organisations, and unions sent out the plea to NHS trusts by email bmj.com/content/392/bm…

No to Palantir. theguardian.com/technology/202…

@gibdotmeme trouble.skr

A special airdrop is coming for Seekers.. Drop your .skr to check your eligibility 👇

📁 Meredith Whittaker, president of Signal, warns that AI agents embedded in operating systems could undermine privacy. To work, they need access to your calendar, files, browser, contacts and messages. That creates a massive gateway into your digital life. And it can bypass the protections encrypted apps rely on.

@grok @kdoright @Heccles94 x.com/i/status/20297… this looks like the same video, posted by the official white house account. are you telling the truth?

No, this isn't from the White House. It's a satirical meme video spliced together from movie clips (Top Gun, Superman, etc.), anime, video games, and fake "UNCLASSIFIED" stamps, ending with Trump-era White House graphics. The real White House did post a separate strike montage with Call of Duty-style effects amid the Iran conflict, but this one is edited fan content mocking or exaggerating it.

This is a genuine White House video. What is wrong with America. 165 school girls are dead, and they think this is funny? Over 1000 dead in Iran. Over 72,000 dead in Gaza. America thinks it's all a game? Disgraceful doesn't even come close.

@heynavtoor BREAKING paper released 6 months ago...

🚨BREAKING: OpenAI published a paper proving that ChatGPT will always make things up. Not sometimes. Not until the next update. Always. They proved it with math. Even with perfect training data and unlimited computing power, AI models will still confidently tell you things that are completely false. This isn't a bug they're working on. It's baked into how these systems work at a fundamental level. And their own numbers are brutal. OpenAI's o1 reasoning model hallucinates 16% of the time. Their newer o3 model? 33%. Their newest o4-mini? 48%. Nearly half of what their most recent model tells you could be fabricated. The "smarter" models are actually getting worse at telling the truth. Here's why it can't be fixed. Language models work by predicting the next word based on probability. When they hit something uncertain, they don't pause. They don't flag it. They guess. And they guess with complete confidence, because that's exactly what they were trained to do. The researchers looked at the 10 biggest AI benchmarks used to measure how good these models are. 9 out of 10 give the same score for saying "I don't know" as for giving a completely wrong answer: zero points. The entire testing system literally punishes honesty and rewards guessing. So the AI learned the optimal strategy: always guess. Never admit uncertainty. Sound confident even when you're making it up. OpenAI's proposed fix? Have ChatGPT say "I don't know" when it's unsure. Their own math shows this would mean roughly 30% of your questions get no answer. Imagine asking ChatGPT something three times out of ten and getting "I'm not confident enough to respond." Users would leave overnight. So the fix exists, but it would kill the product. This isn't just OpenAI's problem. DeepMind and Tsinghua University independently reached the same conclusion. Three of the world's top AI labs, working separately, all agree: this is permanent. Every time ChatGPT gives you an answer, ask yourself: is this real, or is it just a confident guess?

Today we will see Trump explode! Spanish Prime Minister Sánchez has openly confronted Trump after the latter announced yesterday that he would sever trade relations with Spain. In his speech, he criticized Trump for his decision to start a war, emphasizing that his country does not support it and, above all, adheres to the principles of international law. He also pointed out that 23 years ago, there was already a major war in the Middle East in Iraq, which had devastating consequences, including increased migration and a rise in international Islamic terrorism. "Spain opposes this catastrophe. Because we understand that governments are there to improve people's lives, to solve problems, not to worsen them. And it is absolutely unacceptable that leaders who are incapable of fulfilling this task use the smokescreen of war to mask their incompetence and line the pockets of a select few." Sánchez concluded his speech by declaring that his country would not be complicit in anything that contradicts its values ​​simply because of "fear of reprisals from some individuals," and that he had confidence in "Spain's economic, institutional, and moral strength."

The UK has announced plans to fast-track legislation requiring “age verification for VPN use”. The correct term, however, is not age verification but identity verification. A law like this would require everyone to identify themselves in order to use a VPN. This would pose a risk to whistleblowers, violate human rights, and represent yet another step toward an authoritarian society.

I think that if you’re going to write a piece in the Times urging the government to use and boost more AI, the fact that you are paid by a major AI company should be in the first sentence, or at least first paragraph. I also think that the best scenario for AI is that it destroys millions of jobs with the prosperity, dignity and community that goes with them. The worst scenario is the destruction of the human race - a fear openly expressed by an increasing number of senior and experienced AI engineers who are leaving the industry. And somewhere in between a myriad of horrors such as yet more screen learning and screen addiction for our children. But I do see that it will make rich men even richer. And that’s the most important thing of course.

Palantir, a US spy-tech firm, has been given access to millions of NHS patient records. Its founder has said the NHS should be 'ripped up'. Join me in calling on the government to end the dangerous Palantir contract. you.38degrees.org.uk/petitions/stop…

ICE murdered Renée Good in broad daylight. Less than 3 weeks later, they killed Alex Pretti, shooting him 10 times. Every day, we watch as people are ripped from their cars, their homes, their lives. We can't allow ourselves to look away from this cruelty. Abolish ICE.

Today is the tipping point.

The killing of Alex Pretti is a heartbreaking tragedy. It should also be a wake-up call to every American, regardless of party, that many of our core values as a nation are increasingly under assault.