Good morning. Every builder has a trusted method for picking new tools. First, glance at the name. Next, check how many stars it has. Finally, run it and hope for the best. A fake OpenAI repo hit #1 on Hugging Face, pulling 244,000 downloads and 667 likes (from bots) in under 18 hours. The knockoff “Privacy Filter” repo was actually malware built to steal browser data, crypto wallets, and seed phrases. Nothing says “protect my personal info” like a privacy tool that immediately robs you.
Unitree just dropped a giant real life mecha robot. Meet the GD01, a 10 foot tall $650,000 walking robot suit designed for anyone who thinks monster trucks lack subtlety. The launch video opens with CEO Xingxing Wang climbing into its metal belly before it walks, crawls backward like a possessed crab, and smashes through a wall of cinder blocks. Unitree says it’s real and available now, meaning anyone with spare cash and zero impulse control can live out their pilotable robot dreams. Their lawyers even added a “friendly and safe manner” disclaimer, because selling a wall-smashing robot suit tends to keep them up at night. Forget Lambos. China’s answer to the midlife crisis is wearable demolition equipment. (YouTube)
Researchers are trying to teach AI the meaning of life. A new paper from top universities and three frontier labs says AI should help people live better lives. Novel concept. The authors call it “positive alignment,” which means training AI to actively support human flourishing. As AI shifts from answering questions to shaping how people make decisions, the definition of “better” starts to matter a lot more. Cultures, religions, families, and individuals all define the good life differently. The risk is that an AI built to help you flourish might steer you toward someone else’s version of happiness. (arXiv)
The startup playbook is shifting from AI features to full agent workflows. Google for Startups’ Future of AI report is your essential guidebook to understanding how generative media is reshaping product development.
Inside, you’ll discover how to navigate post-keyboard interfaces, multiply output, and create the ultimate defensible moat.
👀 closer look
Wall Street bets on the price of intelligence. CME Group and Silicon Data are launching the first compute futures market later this year, pending regulatory review. The contracts will track daily GPU rental prices, letting traders and cloud providers hedge against compute swings. Why does it matter? Because compute is moving from cloud expense to tradable commodity. CME’s CEO called GPUs “the new oil of the 21st century.” Only GPU pricing is still a mess, with costs swinging wildly depending on who you rent from and what deal they feel like offering that day. If this works, Nvidia gets paid twice: once for selling the chips, and again for manufacturing scarcity. (CME)
Anthropic found a fun new way to piss off OpenAI and Google. Developers and AI agents use APIs to access models, but those APIs can be painful to build with. Dev tools startup Stainless fixes that by turning model APIs into SDKs, the code libraries that help apps connect faster. Now Anthropic is in advanced talks to buy Stainless for at least $300 million. If the deal closes, Anthropic would control the tool developers use to build on OpenAI and Google models. As model quality gets closer, ease of use starts to matter more. OpenAI and Google would have a rival sitting inside their dev experience. (The Information)
AI safety testing is becoming its own security risk. Frontier labs need outside experts to test models before release, but that means giving people access to the models they’re trying to protect. RUSI, a UK security and defense think tank, warns every new access point creates another chance for theft or tampering. That risk multiplies when outsiders are probing models for hacking or bioweapons. The report says the biggest danger is giving testers write access, as they could change how the model responds. Security teams know this problem well. Give people too much access, and someone will find a way to screw it up. (RISU)
Gemini Intelligence is bringing app automation to Android. At today’s Android Show, Google showed off an update that can complete multi step tasks across your phone. Tell Gemini what you want done, and it handles the app work. It can find a class syllabus in Gmail and add the required books to a shopping cart, or use a photo of a travel brochure to find a similar tour on Expedia. Purchases still need approval, because letting AI buy things on your phone without asking is asking for trouble. Google plans to roll it out first on new Pixel and Samsung Galaxy phones before bringing it to watches, cars, glasses, and laptops. Android is starting to look like an AI agent with a home screen. (Google)
fun stats
💥 44%. Global executives who think the AI bubble will burst this year. Nearly 70% are ready to cut AI budgets if the payoff doesn’t show up.
💖 $315,000. Salary for a ‘Claude Evangelist’ at Anthropic. The job requires coding skills, public speaking, and convincing the world Claude is the future. Any takers?
🔥 $160 billion. Potential hit to SpaceX’s valuation if Grok flames out. Downloads of Musk’s AI chatbot dropped by more than 11 million last month.