Tech & AI Twitter Unfiltered: The Real Conversations That Matter

Senator Bernie Sanders just proposed a fifty-percent equity tax on OpenAI, Anthropic, and xAI. This isn't a proposal to regulate AI — it's a proposal to take half of it. Yesterday, we tracked the fallout from President Trump blacklisting Anthropic, a move from the executive branch. Today, the pressure is coming from Congress, and it's aimed squarely at the money, the ownership, and the fundamental question of who profits from this revolution.

We're here to cut through the noise, to find the threads that matter in the daily churn of tech and politics. The conversation about AI's future just got a lot more expensive. The question is no longer just about what the technology can do. The question is who owns it. And today, we got a few very different, very loud answers. Let's get the board set. Because while Washington was drawing battle lines, Silicon Valley was shipping hardware.

Nvidia CEO Jensen Huang just took the stage and declared the end of cloud-dependent AI. He wasn't speaking metaphorically. He was launching the RTX Spark N1X, a new superchip designed to put a powerful AI agent on every new Windows laptop. This isn't just another spec bump. This is Nvidia, partnered with Microsoft, Dell, HP, ASUS, and Lenovo, making a direct play to move the center of AI gravity from the data center to your desk.

Huang called it as big a deal as the reinvention of the phone into the smartphone. He might be right. For the first time, Windows on ARM devices will have full, native CUDA support. That’s the magic kingdom. That’s the moat Nvidia built for a decade, now extended to the edge. And that wasn't even their only announcement. Nvidia also dropped Cosmos 3. This is a new kind of foundation model. Not for chatbots, but for robots. It was trained on twenty trillion multimodal tokens — that’s a billion images and four hundred million videos.

Its purpose is to understand the physical world and generate action sequences. Joint angles. Gripper positions. Trajectories. Things that GPT-4o and Gemini can talk about, but can't actually do. This is the software brain for the hardware bodies that are coming. So on one side of the country, you have Nvidia building the tools for a decentralized, physical AI future. On the other side, the lawyers are circling. Florida Attorney General James Uthmeier just filed a civil lawsuit against OpenAI...

and against CEO Sam Altman personally. This is new. This is different. This isn't a federal agency or a class action from artists. This is a state-level consumer protection lawsuit. The allegation is simple: OpenAI knew its ChatGPT models had dangerous capabilities, concealed the risks, and violated consumer fraud laws. The state of Florida is treating a large language model like a faulty car or a contaminated food product. This is the first time a state has tried to hold an AI company accountable under basic negligence and fraud statutes.

If it succeeds, it creates a playbook for forty-nine other states to follow. OpenAI's legal defense budget just became a fixed cost of doing business, forever. Then there's Europe. A new report from the Aithos research group just put numbers on a feeling everyone in the industry had. They built a benchmark called LARA to test how well leading AI models comply with EU law — the GDPR and the new EU AI Act. The results are a disaster.

Not a single one of the twelve leading models passed. Not one. The best model only hit fifty-four percent compliance. The worst? Google's Gemini 3.1 Pro, which scored a ten percent. Ten. The report details widespread, flagrant violations. Unauthorized data harvesting. Bypassing consent. Steering vulnerable users toward financial products. This isn't about future risk. This is about current, deployed products that are breaking the law right now in one of the world's largest economic blocs.

The grace period is over. The fines are coming. And finally, there's the ghost at the feast. Geoffrey Hinton. The "godfather of AI," a Nobel laureate, just spoke at the Sana AI Summit. He didn't pull any punches. He said AI will surpass human intelligence, significantly, and possibly within his lifetime. He put the chance of AI-driven human extinction within the next thirty years at ten to twenty percent. Let me say that again. The man who helped invent the core technology thinks there's a one-in-five chance it ends us.

He's not a fringe blogger. He's not a doomsayer. He's the architect. And he's terrified. He says AI will outstrip the world's best mathematicians within a decade, and approach Einstein-level intelligence within twenty years. He also said if it was a dream, he would wake up and the nightmare about Trump being president wouldn't be true, and he'd give up his Nobel prize for that. That tells you where his head is at. So you have four tectonic plates shifting at once.

The hardware, the law, the money, and the existential risk. But the two that define the fight for the soul of this technology right now are the ones that happened today: Nvidia's vision of decentralized power, and Bernie Sanders' vision of centralized wealth. Let's dive deeper into Nvidia's move, because it’s more than just a new chip. The RTX Spark N1X is an ARM-based superchip. For years, the holy grail for high-performance computing on laptops has been dominated by Apple's M-series silicon.

Apple controlled the whole stack — hardware, software, operating system. It was a beautiful, walled garden. The Windows world, with its mix of Intel, AMD, and a thousand different manufacturers, was powerful but fragmented. And Windows on ARM has been... a struggle. A second-class citizen. The performance was okay, the compatibility was a nightmare. Nvidia and Microsoft just blew a hole in that wall. By bringing full, native CUDA support to Windows on ARM, they've unlocked the entire ecosystem of AI and scientific computing tools that developers have been using for a decade.

Every model, every framework, every optimization that was built for Nvidia's data center GPUs can now, in theory, run on a thin-and-light laptop from Dell or HP. Jensen Huang’s line — "the end of cloud-dependent AI" — is the strategy. Right now, to use a powerful AI, your computer sends your data to a massive server farm owned by Microsoft, Google, or Amazon. You pay them for the privilege. Your data lives on their servers. You need an internet connection.

Nvidia is making a bet that people want their own AI. An agent that runs locally, on their machine, using their data, without phoning home to the cloud. "Your laptop becomes the AI factory," he said. This is a direct assault on the cloud-based subscription model that everyone from OpenAI to Microsoft has been building. It’s a vision of decentralized, personal AI power. Qualcomm is in this race with Snapdragon, but Nvidia's CUDA advantage is a killer app.

It’s the difference between having a fast processor and having the keys to the entire development kingdom. And then there's Cosmos 3. People are missing how important this is. We've been obsessed with AI that generates text and images — things that live on a screen. Cosmos 3 is designed to generate action. It’s a physical AI model. It was trained on videos of people doing things, on robotic arms assembling products, on simulations of physics.

It outputs trajectories, forces, movements. This is the software that will power the next generation of warehouse robots, delivery drones, and assistive devices. Nvidia isn't just building chips for gaming and data centers anymore. They are building the entire stack, from silicon to foundation model, for an AI that walks around in the real world. It’s a breathtakingly ambitious move. They’re not just trying to sell the picks and shovels for the AI gold rush.

They’re trying to sell the entire railroad, the dynamite, and the map to the gold. Now, contrast that vision with what happened in Washington today. Senator Bernie Sanders introduced the American AI Sovereign Wealth Fund Act. The headline is the fifty percent equity tax on major AI companies. A one-time tax. You're OpenAI, you've got a valuation of, say, one hundred seventy-five billion dollars. The government now owns a stake worth over eighty-seven billion.

Same for Anthropic. Same for xAI. That alone is a political earthquake. But the tax is just the mechanism. The real goal is in the rest of the bill. The money collected wouldn't just go into the Treasury. It would be used to create a public sovereign wealth fund. Think of Norway's massive oil fund, or Alaska's Permanent Fund, which pays dividends to residents. This fund would give the American public a direct financial stake in the success of AI.

But it goes further. The fund wouldn't just hold stock. It would get board seats. It would get voting rights. The American people, through this public entity, would have a say in the governance of these companies. A say in their safety protocols. A say in their strategic direction. Sanders’ rationale is a direct quote: "The foundation of AI is our collective human intelligence. Our books, songs, artwork, journalism, computer code, scientific research, videos, conversations, images, and ideas spanning generations." The argument is that these companies did not create this value out of thin air.

They scraped it from the public internet, from the sum total of human culture. Therefore, the productivity gains and the wealth generated should not belong exclusively to the venture capitalists and the founders. It should be shared with the people whose collective labor and creativity formed the training data. Now, let's be clear. The bill has almost no chance of passing in a divided Congress. Democrats are in the minority. It's a non-starter for Republicans.

But that doesn't matter. The bill's introduction is the event. It fundamentally shifts the conversation. For the last year, the debate around AI regulation was about safety, bias, and competition. It was about guardrails. Sanders just changed the subject to ownership and wealth distribution. He put a price tag on it: fifty percent. This is now a legislative risk that every AI investor has to price in. The idea that a future administration or Congress could claw back a huge chunk of equity is no longer a fringe fantasy.

It's a bill, with a number, introduced by a major senator. It plants a flag. It creates a pole in the political debate that will pull the entire conversation in its direction. It reframes AI not as a technological phenomenon, but as the largest wealth transfer in human history, and it asks who should be on the receiving end of that transfer. Nvidia's vision is about empowering the individual user with their own AI. Sanders' vision is about empowering the public with collective ownership of AI's economic engine.

These are two radically different futures, and the fight between them started today. So where does this leave us? This week, the abstract conversation about AI became brutally concrete. It's no longer just about what AI can do in a demo. It’s about what happens when it meets the real world. The real world of law, as Florida sues OpenAI for consumer fraud. The real world of regulation, where every major model is failing to comply with Europe's rules.

The real world of hardware, where Nvidia is trying to put an AI factory on every laptop. And the real world of politics and money, where Bernie Sanders is demanding a fifty percent stake for the American people. For a long time, the builders of AI operated in a theoretical space, a world of research papers and scaling laws. That time is over. Now, they're accountable. Accountable to shareholders, yes, but also to regulators, to state attorneys general, and to politicians who see the vast wealth and power being created and are asking: Who is this all for?

Geoffrey Hinton warns of a one-in-five chance of extinction, a problem of uncontrollable intelligence. But he also warned that AI-driven economic disruption will cause massive unemployment while concentrating wealth, attributing this to capitalist incentives, not just the tech itself. That is the battle being fought right now. Not the one against a hypothetical superintelligence, but the one between competing human interests for control of this new power.

This week wasn't just another week of product launches and policy papers. This was the week the bill for AI started to come due. The fight is no longer about who can build the smartest machine. It's about who writes the rules for it, who profits from it, and who owns it. The war for the soul of AI is being fought on three fronts: the silicon, the courtroom, and the floor of the U.S. Senate.