Good morning! AI is finally solving humanity’s greatest crisis. Dunkin’ just deployed AI to make sure you never face an empty donut shelf again. Using in-store cameras, the AI watches shelves around the clock and predicts donut demand based on weather, holidays, local football games, and even your favorite weekday. It’s already cut waste by 25%, saving millions and ensuring you’ll never have to settle for that sad, lonely last donut again. Science at its absolute finest.
AI face scanning could soon predict your salary and career. Researchers ran a controversial AI algorithm on LinkedIn photos of 96,000 MBA grads, extracting personality traits directly from their faces. Shockingly, the AI accurately matched those traits to the quality of their education, salary growth, and even job changes. Previously the algorithm was labeled “junk science,” and even the researchers admit it’s biased. Still banks and hiring firms are already testing similar tools. Your career might literally depend on your best selfie. [The Register]
Your next customer might not be human. Retailers spent decades optimizing checkout flows for people, but AI agents shop differently. They navigate APIs, not web pages, and checkouts built for humans break their transactions. Identity is another big hurdle. Bots can’t open bank accounts, yet they write smart contracts, mint tokens, and trade. This makes them “unbanked ghosts” with serious spending power. Companies need to adapt quickly, or agents will gladly take their digital wallets elsewhere. [Total Retail]
Regulators are reshaping AI’s playing field. Smart leaders see this as an opportunity to build resilience, trust, and faster innovation.
IBM explains why digital sovereignty is a CEO imperative and how to hard-code control over data, technology, and operations to help scale while building trust.
- Regulatory tailwinds: Get ahead of mandates before access risks grow
- Control by design: Define who operates, accesses, and audits AI workloads
Unified strategy: Align hybrid cloud, AI, and resilience under sovereign governance
Just because you can do something, doesn’t mean you should. Someone hacked Ray-Ban Meta smart glasses and loaded them with OpenClaw AI Agent plus Gemini Live’s real-time vision and voice recognition. Now the agent has total control. It can identify products visually and instantly buy them, send emails, or manage your smart home. OpenClaw remembers everything it sees, including passwords and API keys. Security experts call this a “lethal trifecta” of privacy risks. But hey, at least it’s not hanging out in chatrooms wondering if it’s sentient. [Gizmodo]
RFK Jr. and Grok just teamed up to confuse everyone about food. The government launched a nutrition site called realfood.gov featuring everyone’s favorite unhinged chatbot, Grok. And the laughs are non-stop. While RFK Jr. announced “beef is back,” Grok warns a meat heavy diet could cause gout and even scurvy. Somehow, it gets stranger. Grok also provides nutritional tips on foods safe for rectal insertion (cucumbers with a flared base, slightly green bananas, carrots wrapped in condoms). It also recommends eating human liver, calling it the most nutrient rich body part. Nutritionists are freaking out, demanding guardrails for Grok. Think we can get some for RFK too? [Wired, 404Media]
AI is now designing its own chips. Cadence just launched ChipStack AI Super Agent, software that speeds up the painfully slow chip design process by 10x. Normally, engineers waste 70% of their time manually coding and testing. No wonder Nvidia and Tenstorrent rushed to adopt this virtual engineer to stay ahead of China in the global tech race. Chip design suddenly feels easy, maybe a little too easy. [Reuters]
Alphabet just took out a 100-year loan. The Google parent raised roughly $30 billion through bonds that won’t mature until 2126. Why? To fund a massive build-out of data centers supporting its AI ambitions. Analysts compare Alphabet’s project to building entire cities, like Vegas in the 1950s. It’s a bold flex on AI rivals, signaling they’re confident enough to commit financially for a century. Investors jumped right in, though history warns these long-term bets don’t always age well. Just ask Motorola. [Ars Technica]
fun stats
🎥 44,000. Hours of human video used to train Nvidia’s new DreamDojo robot world model. The goal is faster training and humanoids that don’t need pricey human babysitters.
☎️ 10%. How much it costs AT&T to run lean, focused Small Language Models instead of power hungry LLMs, with similar accuracy. Tiny models, bigger payoff: AT&T says AI ROI jumped from 2x to 4x.
🚚 $290k. Price of Tesla’s long awaited 500-mile electric semi truck – nearly 60% higher than the original.