The Artificial Intelligence

A 24-hour coffee shop run entirely by a single robot opened in Beijing on Sunday, offering coffee with an average wait time of less than 30 seconds per cup. On its opening day, the shop served 202 cups of coffee in one hour, setting a Guinness World Record for the most cups of coffee served by a robot-operated store in one hour.

Anthropic CEO Dario Amodei's wife and adviser, Cami Clark, pitched Jeffrey Epstein on investing in her "free luxury porn company" in 2012, per DOJ-released emails. Epstein, then already a registered sex offender, declined: "Can't do sex TV." Dario Amodei…

Anthropic CEO Dario Amodei's wife and adviser, Cami Clark, pitched Jeffrey Epstein on investing in her "free luxury porn company" in 2012, per DOJ-released emails. Epstein, then already a registered sex offender, declined: "Can't do sex TV." Dario Amodei and Cami Clark weren't married at the time, just to be clear. They started dating in 2014.

Gemini 3.7 Flash is one point behind GPT-5.6 Terra on intelligence and 40% faster per task. It scores 56 while generating ~340 tokens/sec and costing about $0.40 per Artificial Analysis task at Google's introductory price. Google is getting VERY aggressive with the intelligence-per-dollar-per-second thing.

OpenAI CEO Sam Altman: "in the next 6 months, we're close to a world where a descendant of ChatGPT can watch your screen, record every meeting and call, and have perfect context of your whole life" We're only one model generation away from this being incredibly useful

AI Is Slowly Poisoning Itself (And No One Can Stop It) https://youtu.be/lgWLuwQtGs8

Square Enix is working with Google to use AI for testing video games, the project is also backed by Japan’s Ministry of Economy, Trade and Industry. The company has introduced an AI-powered QA system built with Google Gemini, the AI can play a game, understand on-screen events, listen to sounds, check the map and control the character while searching for problems. They want to automate repetitive testing tasks such as finding bugs, checking graphics and verifying text. Square Enix says the technology could reduce the work for developers and testers, freeing them to focus on other aspects of game development.

Tech journalist Ed Zitron asks the trillion dollar question: what happens to Microsoft and Google the moment OpenAI stumbles?

Ilya Sutskever’s company is going to release a model in August. I took another look at their website. They state it very clearly: their only goal, their only mission, is to develop superintelligence. That is their sole focus. In that sense, this model release will be far more than “just” another model release. In some form, they must have achieved superintelligence if we take their statement seriously. "SSI [Safe Superintelligence] is our mission, our name, and our entire product roadmap, because it is our sole focus. Our team, investors, and business model are all aligned to achieve SSI."

Palantir CEO: "I've spent a lot of time with Dario and the effective altruism crew - they want to tell you: we have to march into a future where we own nothing, where our businesses aren't profitable, where none of us have jobs" Alex Karp just called the frontier labs parasites and ended the interview when Sam Altman came up

Google Turned $900 Million Into $94 Billion (And Can't Touch It) https://youtu.be/nk7WFTc1ugU

Qwen 3.8 Max just built better 3D physics scenes than Fable 5 while costing about 7x less to run. Test was done by atomic chat, a desktop app that runs LLMs locally. Outputs: Qwen 3.8 Max: 46.6K tokens, $0.28 Fable 5: 38.7K tokens, $1.93 Both models got the same brief, three self-contained HTML files, each a scene that simulates itself. The prompts asked for a marble machine lifting marbles onto a wheel and a loop, a car assembly line, and a sawmill cutting logs into planks. These tests were actually pretty brutal. Shows whether the model maintained object state, collisions, and machine geometry across the whole scene.

TPU 8i and TPU 8t deliver sustainable economics—offering up to 80% better performance-per-dollar for low-latency serving while seamlessly scaling to 1M+ chips!

Wow. This futuristic idea is getting very closer to reality. 👀 "Future space stations may no longer be built by astronauts." > MIT Media Lab and the Aurelia Institute just unveiled their latest vision for TESSERAE, a self-assembling space habitat system that uses intelligent magnetic tiles to autonomously snap together in orbit. > Inspired by geodesic domes, each tile contains electro-permanent magnets, sensors, onboard electronics, and distributed control software, allowing hundreds or even thousands of flat-packed modules to assemble into large, reconfigurable habitats far bigger than a single rocket could launch. > Since its inception at MIT in 2016, TESSERAE has progressed from laboratory research to parabolic zero-gravity flights, Blue Origin's New Shepard, ISS demonstrations aboard Axiom Mission 1 in 2022, a full-scale habitat pavilion in 2024, and showcases at TED2025 and MIT Space Week 2026. Team is now working toward Aurelia's first orbital self-assembling habitat, bringing one of the most ambitious concepts in space architecture closer to reality.

Nvidia is making a “substantial” investment in Ilya Sutskever’s Safe Superintelligence, and providing enough GPUs to increase its compute tenfold. SSI had previously relied mainly on Google’s TPUs. The new long-term partnership gives the secretive AI lab access to Nvidia’s next-generation Vera Rubin platform. Nvidia made the investment after receiving a rare look at SSI’s research. Financial terms were not disclosed.

Jensen Huang second post ever: he isn’t just arguing for open AI anymore but building an alliance around it. NVIDIA has launched the Open Secure AI Alliance with companies like Microsoft, Hugging Face, Palo Alto Networks, Salesforce, SAP, Red Hat, Cloudflare and many more. The message: attackers already have frontier AI. Defenders need an open ecosystem to keep up.

🇺🇸🇨🇳 Apple wants to use Chinese memory chips to lower the prices of their products, Micron the only major U.S. memory chip maker, is lobbying against, per WSJ Apple is lobbying the Trump administration to allow it to use memory chips from China's ChangXin Memory Technologies and Yangtze Memory Technologies in products sold outside the U.S., arguing the move would ease the global memory shortage and help lower costs, WSJ reports. Micron, the only major U.S. memory chip maker, is lobbying against the plan, warning it could undermine the domestic semiconductor industry, even if the chips are only used in products sold abroad.

Apple wants Trump to approve Chinese memory chips; Micron warns the move could destroy America’s last major memory-chip producer. Via WSJ Tim Cook has reportedly asked the administration to let Apple use chips from China’s CXMT and YMTC in products sold outside the US. Apple says this would ease the global shortage and lower costs. Memory-chip prices have quadrupled over the past year, while Micron’s gross margins now exceed 80%. Apple accuses the company of raising prices and reserving too much new capacity for AI data centers. It's crazy to see that China is increasingly becoming the supplier of almost everything.