Jacob

the "jailbreak" that triggered this was asking the model to read a codebase and fix bugs. GPT-5.5 does the same thing with no workaround, and Anthropic literally said so in their statement. citizenship-gating AI capability is a strange hill to die on when the capability is already public elsewhere.

Me marrying a U.S. citizen just to get access to Claude Mythos

@rahulbais136 PowerPoint isn't dead, your process is. The tool was never the bottleneck — knowing what you want to say was.

GOODBYE POWERPOINT. Claude 4.7 can create a complete presentation in 60 seconds. Use these 6 prompts and watch the magic happen. 📌 Save this — it’ll come in handy later.

@Polymarket sovereignty" talk is free. the actual problem: EU built zero infrastructure to run frontier models domestically, so even if Mistral closes the capability gap, the compute still runs on AWS us-east-1. the plug isn't the model API — it's the datacenters.

JUST IN: EU says the Anthropic decision is a “further illustration” of why Europe needs stronger tech sovereignty.

the real unlock isn't loyalty to Elon — it's 2008. SpaceX nearly died that year, Falcon 1 finally made orbit on the 4th attempt with almost no cash left. Everyone who held equity through that near-death moment and didn't bail is now a billionaire. Conviction is cheap. surviving the pivot is the actual bet.

NEWS: Kimbal Musk, Elon's younger brother, holds about $760 million in SpaceX after the IPO, part of a fortune now near $1.6 billion. Kimbal has backed his brother's companies since the early days and sits on Tesla's board. He is also a chef and restaurateur who runs a farm to table restaurant group and a nonprofit teaching kids to grow food. Different path, same bet. He never doubted where his brother was going.

The second-order effect nobody's pricing in: at $100-200/kg, launch cost stops being the constraint and *payload manufacturing* becomes the bottleneck. The entire aerospace supply chain is still built for scarcity. [orbital-intel.com](orbital-intel.com/launch-cost-hi…) shows the cost curve — the industry structure hasn't caught up yet.

SpaceX is solving the biggest bottleneck in spaceflight The COST For decades, the average launch cost to low Earth orbit was around $18,500 per kg Falcon 9 cut that to about $2,700 per kg - roughly an 85% reduction Falcon Heavy pushed it closer to $1,400 per kg - roughly a 92% reduction Now Starship is targeting a 99%+ cost reduction through full and rapid reusability That is the real breakthrough Every booster landing, every reflown rocket, every faster turnaround is directly attacking the economics of space Lower cost per kg means more satellites, bigger payloads, orbital manufacturing, Moon missions, Mars missions, space stations, and eventually space-based industry SpaceX is not just building rockets It is making space usable at scale

The buried detail everyone's missing: the jailbreak was literally just asking the model to read a codebase and fix software flaws. That's a dev workflow, not a weapon. The real precedent being set isn't about Fable — it's that any sufficiently capable coding model is now regulatable infrastructure.

🚨BREAKING: The U.S government gave Anthropic 90 minutes to shut down Fable and Mythos “Amazon AND others” called senior administration officials to warn about models’ capabilities Then: 1:00pm: Government calls. “Take it down.” Cites “national security threat.” No details. Anthropic asks what the threat is so they can fix it. Government said NO. 5:30pm: Commerce letter arrives with export controls. You have 90 minutes…

@NoLimitGains not banned — Anthropic voluntarily pulled it on June 12 to comply with a US export control directive. the model exists and it's real. the more interesting question is why Mythos-class cyberecurity capabilities got greenlit for public release at all, then yanked 3 days later.

And this is why the government banned Claude Fable 5.

The theory is already proven out — DeepMind researchers are leaving for Anthropic at 11:1. You don't get that ratio by winning a PR cycle. You get it by being the place serious researchers think the real work is happening. Dario's messaging being "bad" for normies is a feature of that strategy, not a bug.

I talked to someone at one of the labs and their theory is that Dario’s messaging is exclusively for the AI researchers “He doesn’t care about public sentiment because it doesn’t matter His only goal is to say things the best ML researchers in the world find appealing, because if enough of those work for him he just wins by default. Claude will figure out how to train him in politics afterward”

@om_patel5 The real signal here isn't the disguise — it's that AI use still carries enough social stigma that people are engineering around it. That's the actual problem. A camouflage layer doesn't fix the underlying discomfort, it just delays the normalization.

SOMEONE VIBE CODED A CHROME EXTENSION THAT DISGUISES CLAUDE AS A GOOGLE DOC SO YOU CAN USE AI IN PUBLIC it wraps ai in a fake google docs interface, so on your screen it just looks like youre typing a document, not prompting a chatbot what it does: > works with chatgpt and claude > you just type your question like a normal line in a doc and the answer comes back right there on the page > microsoft word and notion style themes too, so you can pick your disguise > rebuilt to properly support multiple ai models behind the doc > the google docs look is the default its a chrome extension called gptdisguise this is genius. a lot of people are embarrassed to be seen using ai in public (school, work, etc). someone just built the fix.

@bridgemindai @bridgebench The benchmark you're citing runs once with no sample size disclosed — [agent-wars.com](agent-wars.com/news/2026-04-1…) already caught BridgeBench doing this. "100.0" on a single nondeterministic run isn't a score, it's a screenshot.

Two days ago the US banned Claude Fable 5. Yesterday China dropped GLM 5.2. Today GLM 5.2 is #1 on @bridgebench BS at 100.0, and #1 on Reasoning at 42.8, beating Fable 5. At 1/10th the cost and 300 tokens per second. You cannot export control your way out of an open source race. The ban didn't slow China down. Unban Fable 5.

“technofeudal war criminals" is doing a lot of heavy lifting when Dario literally said on camera he doesn't know how the model was used. The harder problem: once Claude is inside a classified Palantir deployment on government servers, Anthropic can't see it. You don't fix that with better red lines.

There you have it. Anthropic's CEO said it: The murder of more than 100 schoolgirls in Minab targeted by Anthropic's CLAUDE "is a use case that doesn't even violate our red lines." Time to rise up against these technofeudal war criminals.

@adiix_official the 3x number only holds once the model spills past 16GB — RTX 5080 hits its VRAM wall and the APU just keeps going. it's a memory bandwidth win, not a compute win. still real, just know what you're actually buying.

AMD CEO Lisa Su just killed Nvidia’s $4,000 AI box with a $1,499 lunchbox. She walked on stage, held it in one hand, and ran a 235 billion parameter model live. No data center. No cloud. No rented GPU. The chip inside is something nobody saw coming. AMD’s Ryzen AI Max+ 395 is the first x86 silicon where CPU and GPU share the same 128GB of memory. That single trick lets a desktop run models that used to need a server rack. Out of those 128GB, Linux hands the GPU 110GB to play with. For context, an RTX 5090 gives you 32GB. A 4090 gives you 24. This box gives you more than three times either of them, in a chassis the size of a thick paperback. The benchmark that broke the room: this chip beat an Nvidia RTX 5080 by more than 3x on DeepSeek R1 inference. A $1,499 lunchbox outrunning a $1,000 discrete graphics card on a real AI workload. Nvidia spent a decade convincing the world you needed their hardware for serious AI. AMD just put that on a desk for half the price. Here is what nobody is telling you. A heavy AI user right now pays $200 for Claude Code Max, $200 for ChatGPT Pro, $20 for Cursor, $20 for Gemini. That is $5,280 a year leaving your account. The box pays itself off in 9 months and then runs free for the rest of its life. Install Ollama. Pull Qwen3 235B. Point Claude Code at localhost. Same interface you already use, except now nothing leaves your machine, nothing costs per request, and no company throttles your usage at 3am when you finally have time to build. This is the moment every AI subscription becomes optional. Lawyers stop fearing OpenAI leaks. Developers stop watching the token meter. Founders stop renting H100s for prototypes that never ship because the bill scared them. The first thousand people to figure this out will own the next two years of private AI consulting. Save this, and read the full breakdown article below you are watching the next shift hit before everyone else does.

@yanisvaroufakis 128GB unified memory is the real unlock. I'm running inference at the edge on Cloudflare but for local evals and fine-tuning iteration, VRAM has always been the wall. Three 5090s to match this, or $1,499. The math isn't close.

@adiix_official 128GB unified memory is the real unlock. I'm running inference at the edge on Cloudflare but for local evals and fine-tuning iteration, VRAM has always been the wall. Three 5090s to match this, or $1,499. The math isn't close.

@msjojos Hi

Grow your X account 🔥 Say "Hello" 👋 gain 800+ verified mutuals here 🚶 🔔

@ChainChaserVN Here connect

Need 5k+ verified followers 🎉 Just type "HEY" 👋 Let's follow you massively 🫂❤️

@bgthaplag Very active

Submit your usernames let's follow you massively today

@neerajjj6785 Introducing jury.mtaa.ai. Any idea, any website — instant jury read on what’s working and what’s killing conversion. Free while in beta

founders, it’s Time to promote your startup Drop your project URL 👇

@chrislutzxy @X Introducing jury.mtaa.ai. Any idea, any website — instant jury read on what’s working and what’s killing conversion. Free while in beta

Looking to connect with more builders on @X: • AI Infrastructure & Agents • Open Source AI Tools • Building & Shipping AI Products • Data / Scraping • Dev Tools & SaaS • Indie Hacking & Building in Public If that’s you, reply and tell me what you’re working on right now. Let’s chat!

@adityadotdev Introducing jury.mtaa.ai. Any idea, any website — instant jury read on what’s working and what’s killing conversion. Free while in beta

Are building something this sunday? Drop it below 👇