Nishiibe built a browser game in three.js where a low-poly kappa, the water creature from Japanese folklore, goes through a full sauna routine. Per the creator, the goal is to get the kappa balanced in the sauna. In the clip, the kappa in sunglasses and a towel moves from the hot room to a cold plunge to a lounge chair outside, the screen bursts into aurora and kaleidoscope effects, and a score card compares each step with its target time, ending at 99 points. Why it matters: It turns a real ritual, the Japanese sauna cycle of heat, cold water and rest, into a simple timing game anyone can play in a browser, and shows how much charm a small web 3D project can carry. Credit: @nishiibe https://t.co/EPkjIAjzTp

“The times are never so bad that a good man cannot live in them.” — Saint Thomas Aquinas

“The times are never so bad that a good man cannot live in them.” — Saint Thomas Aquinas

It may have taken just one compromised laptop to steal over $6 billion from Binance. The biggest crypto hack ever. Binance says it will cover the losses. But if even the largest crypto exchange can be hit this hard, how safe is the rest? https://t.co/PRgqtewD9Q

An Accenture worker reportedly gave a personal AI agent access to a secure Pentagon system. The agent may have shared passwords and sensitive data. Nobody noticed for two months. AI agents can do a lot. But we must be careful what we let them access. https://t.co/BnQ5rQkpSX

By https://t.co/AXFDMteTsh via @wired https://t.co/vkPUJFyJ2n

In this new paper, we make a strong point: Human imitation is a misplaced goal for AI. Too much of AI development is based on human imitation, as exemplified by the Turing test: the more a machine resembles us, the more successful it is. But there are two major problems. The first is conceptual. Imitating human outputs does not tell us much about the capacities behind them: The same answer can emerge from different processes. A plausible response does not tell us whether a system can reason reliably, generalize to unfamiliar situations, or recognize when it should say “I don’t know”. The second is ethical. Why should we want AI that imitates humans? Humans are biased. Our attention is limited. Our judgment is fallible. Reproducing these limitations is hardly an ambitious vision for AI. Even worse: A system that impersonates a person can exploit human trust. And an industry focused on replicating humans can fall in the “Turing’s trap”: favour human replacement over augmentation. Instead, we argue that AI should aim to improve human welfare. To this end, we propose three coordinated changes: For developers: Build systems that strengthen human capabilities and independent judgment. For funders: Require projects to assess benefits, costs, and who receives or bears them. For institutions: Base procurement and renewal on demonstrated benefits. The goal should be to help people know more, suffer less, and exercise greater control over their lives. * Full paper in the first reply. With @GaryMarcus & @Walter4C

Steve Jurvetson once told Elon Musk a story he had never heard before. At a Google event, Larry Page turned to Jurvetson out of the blue and wondered aloud how much money he had. After mentioning a number, Page said that if he were hit by a bus that day, he should leave all of it to Elon Musk. Elon asked if he had really said that. Jurvetson confirmed it and joked that he immediately wanted to get it in writing. Elon said he loved that and called Page a good friend. He noted they had known each other since around 1997, before Page received venture funding. Jurvetson added that Page believed a corporation pursuing goals aligned with his own interests—such as moving away from oil and colonizing other planets—would accomplish far more than giving the money to a nonprofit. Page wished he could do more of that himself. https://t.co/rn5rE4z882

"I am a citizen of the United States and have only that passport. No matter what happens, I will fight for and die in America." Elon Musk https://t.co/hbzWiJR5Qm

Totally blind, first ride in Robotaxi, the app was totally accessible, the ride accessible, the drop-off accessible, it felt unusually normal! @TeslaSI https://t.co/l4UQTXB66J

I still think this needs to be hung in the Louvre https://t.co/U0Nlk5QN0P

Obama: End the filibuster and make Washington DC and Puerto Rico states. You are crazy if you don’t believe Democrats will do it if they get the chance. https://t.co/w6cXLwWbWb

Our picks from this week's model releases 🦙 → LFM3 by Liquid AI: small models for text, image and speech, with d1 decision models → Mistral Large 4: large, 1M context frontier → Flint by Recursive: mid-sized, built for coding agents Get them at https://t.co/GP9txkNJEd https://t.co/yqloGmelr7

That’s a wrap on @MusicTectonics! The Stable Audio team spent two days in Santa Monica showing how you can create with our tools, meeting people from across the industry, and joining the conversation on where AI and music go next. https://t.co/BO4wvy1MLl

Our CEO @premakkaraju sat down for a chat with Roland’s @mccabep on building trust in AI tools for music creation, and our SVP of GTM @AnaGuillen joined @ElevenLabs' Bridget Ferris and Royall’s Anne Kavanagh to discuss how technology can grow the economic opportunity for music. https://t.co/qv0AVUWRy3

Thanks to everyone who stopped by! If you didn't catch us there, we'd love to chat: https://t.co/Pc584nXDVB https://t.co/om6k9CWUT7

+1 more

Thanks to everyone who stopped by! If you didn't catch us there, we'd love to chat: https://t.co/Pc584nXDVB https://t.co/om6k9CWUT7

The @AbridgeHQ Assistant is the AI assistant for healthcare. Our vision is that every clinician has a team of agents figuring out all the things they want to know, helping them ask the right questions, leverage the right evidence, and improve patient care and outcomes. Now in use at 300+ health systems.

@celineodier My agent is better. Reads 30,000 posts a day in AI community here on X and builds this: https://t.co/OSeoAG4VSQ

Figure has paid regular people $15 million to film themselves doing chores, and that data made its robot 6x better in homes it had never seen. it's an app called Index. you sign up, get a recording device and get paid by the minute while you fold laundry, make beds or clean a kitchen the numbers from Figure's own post: - 264,000 downloads in 108 countries - 44,000 people uploading every week - 16M+ videos, 30 minutes of footage every second, which Figure counts as 4.9 years of human work per day - $15M paid out so far, and over $1B committed to data and compute in the next 12 months then Figure pretrained Helix 2.5 on it and sent the robot into 30 Bay Area homes it had never seen, with zero data from any of them same task data, same evaluation: zero-shot success went from 9% without Index to 56% with it and Figure isn't the only one buying human hours: - NVIDIA trained GR00T N1.7 on 20K hours of EgoScale, people filming their own hands - Tesla hired data collection operators at $25–48/hr to wear motion-capture suits and VR headsets for Optimus - Mecka raised a $60M Series B, reportedly led by Sequoia, to pay people to record everyday tasks with body sensors main insight: two years ago, teaching a robot to fold a towel meant a lab, a teleop rig and someone with a PhD now the fastest way is paying a stranger to fold their own towel on camera, 44,000 strangers a week

Free image-to-3D models don't preserve your source image - text, logos and fine detail come back garbled. I got that working with my latest 'Pixel Match' feature in my image-to-3dlab: it copies the actual pixels from your source onto the model, even after retopo down to ~5k faces. High quality, low poly 3D models that are riggable and can be animated. One step closer to Tripo / Meshy level of output, but free and on your own machine. Mac + NVIDIA. Try it out here: https://t.co/fCh7Ft9x9m There's lots more to do, so please test, raise issues & PRs, etc.

I’ve had so much fun making bricky, and playing with it: a compiler for LEGO® models, built for AI agents. The agent writes a Python build script. bricky places every part, reports each mistake, and renders the model. Thanks to @omar_aflak for the idea of a LEGO® compiler. https://t.co/9qKjNzTbhf

Introducing Stepers : Email marketing for SaaS, built for the agent era 👋 Manage your entire email stack from one place. → Transactional emails, workflows → Build and manage everything with AI agents → Connect your SaaS and turn user events into automated emails → Migrate from your current provider in minutes Your entire email stack, powered by AI. Use it with your favorite AI agent. Your email copilot is here. Try it here : https://t.co/33btrf33wl

Tasklet now has an iOS app! Delegate work to your team of AI agents, wherever you are. Put them to work across your tools and the web, set up automations, build custom dashboards, and collab with your team. All from your phone. Here's what you can do with the app 🧵 https://t.co/CLhpo5giUM

Regarding autoresearch agents, this question has attracted a lot of attention. Do research agents have taste? This has attracted massive interest because autoresearch agents could unlock major discoveries. This study gave six frontier models seven recent NLP papers, then ran every experiment they proposed, including the ones they ranked last. Of 415 experiments, 194 were judged sound, valuable, and new. Only 73 were also beyond the seed paper and non-obvious. The rankings added little. First choices beat a random pick, but they were not shown to beat the experiment written first. Forecasts beat a base rate, but they were not measurably better than a fresh instance of the same model that read only the paper and registrations. They tested with teams of agents, which cost more, with no clear improvement in proposal quality. https://t.co/wF4ajNDptV Chat with Paper: https://t.co/uokhb72MyB

Privacy should advance as fast as intelligence. OpenAI Private Intelligence gives businesses stronger controls over their data while using our most capable models. We’re building these protections in close collaboration with some of our largest customers to help businesses apply frontier AI to their most sensitive work. https://t.co/BiA4N8FQVH

Computer's Mac and Windows apps now work in your own Chrome. The new Perplexity extension links the two, so Computer uses sites you're already signed into. Here, it turns a business's reviews into a prioritized list of fixes and replies. Live now for all Computer users in the Mac and Windows apps. Get it: https://t.co/IuPEEK6bZ0

Perplexity Computer can now control your Google Chrome browser from the Perplexity Mac and Windows app. A new chrome extension below to set it up once and you can just ask your agent to do any local browsing tasks on your Chrome browser https://t.co/YBtzQ7NYLP

Tibo, a freelance technical artist, shared new swimming tweaks for Wild Cosmos, a game being built in the Godot engine. Per the creator, the update smooths the transition between the ground and swim states when the character enters or leaves the water, plus minor visual effects and sound changes. In the clip, a small character in a white helmet glides through a bright blue sea that curves around a tiny round world. Why it matters: Transitions like this are easy to overlook, yet they decide whether movement feels smooth or clunky, and it is nice to see that kind of polish shared openly while the game is still being made. Credit: @heytibo https://t.co/LN9y0kBF6g

@thdxr Been doing this for a while... my own client talks to Claude Code through the ACP adapter, on my regular subscription. Claude Code still does the work, I just pick the front end. That split is the whole point of ACP. https://t.co/h0kfQWJEMA

Hey New York 🗽 It's time to push the boundaries of what's possible with Gemini 4 💎 @cerebral_valley and @GoogleDeepMind are teaming up to bring you: 🏆 $20,000 in prizes 📍 In-person in New York City 👥 Teams up to 4 Space is strictly limited. Apply now 👇