Good morning. Smart glasses are undeniably cool. Instant videos, built in AI, and facial recognition. What’s not to love? Oh right, privacy. Luckily, someone built an app to let you know if you’re starring in someone else’s livestream. It’s called “Nearby Glasses,” and it scans Bluetooth signals to warn you when smart glasses are nearby. The developer calls it a “tiny act of resistance” against surveillance tech, but mostly it’s about not becoming an accidental influencer.
The Pentagon is threatening to seize Anthropic’s AI. Dario has until Friday to accept the military’s demands, or the government will invoke the Defense Production Act and take control of Claude. Anthropic has a $200 million contract integrating Claude into classified defense systems, but Dario draws the line at autonomous weapons and mass surveillance. Last week, defense officials threatened to label Anthropic a “supply-chain risk” before realizing they’d shot themselves in the foot. Blacklisting Anthropic would force contractors to pull Claude, crippling military AI projects and leaving everyone scrambling for replacements. And replace it with what exactly? Grok? Deepseek? Anthropic isn’t backing down on safeguards, but maybe getting banned by the Pentagon is just good marketing. [WP, Reuters]
Prediction markets jumped to a 42% chance this won’t end well for Anthropic.
Economists can’t decide if AI actually boosts productivity. AI was supposed to kick off a productivity boom, and at first glance, the numbers looked promising. GDP grew 2.2% last year while job growth barely budged at 0.1%. The problem is that nearly all that growth came from pouring billions into AI data centers, not from people working smarter. Half the workforce claim they use generative AI, yet only 13% actually use it daily. Individual tasks might speed up by 40%, but overall productivity rose just 0.5%. Still, companies restructuring around AI agents are seeing stronger results. Economists can keep arguing, but one thing’s clear: if AI doesn’t start delivering real productivity soon, that huge price tag gets tough to defend. [Economist]
Ditch your keyboard. Wispr Flow turns your messy um speech into clean, polished text. Works in every app, 4x faster than typing.
LinkedIn’s former CEO got “voice-pilled” and now sends 89% of his messages with zero edits using Flow. So do millions of others, including teams at Clay, OpenAI, and Vercel.
– Speak to Claude, Slack, Gmail, and anywhere else you type.
– Flow understands code and context.
– Talk and you’re ready to send.
Now on Android, and free during launch. Try Wispr Flow.
Musk just lost his trade secrets lawsuit against OpenAI. A US district judge tossed out claims that OpenAI stole its “secret sauce” by hiring eight ex-xAI employees. While some of them may have walked out with source code or private chats, Musk presented zero proof OpenAI orchestrated it. It’s the latest chapter in Musk’s ongoing feud with Sam Altman, complete with lawsuits, public clashes, and endless drama. OpenAI called the whole thing “baseless” and celebrated by dismissing the suit as another Musk “harassment campaign.” [Verge]
Smart factories now predict breakdowns before they happen. Factory downtime burns about $260k an hour, turning into a $1.4 trillion headache every year. Now investors are shifting cash from heavy machines to smarter software designed to eliminate downtime altogether. AI powered factories don’t just predict potential issues. They use live IoT sensor data to pinpoint exactly what’s going wrong, transforming maintenance from guesswork into clear, targeted fixes. And these aren’t massive, complicated systems. They’re agile cloud solutions that factories roll out in weeks, rapidly turning downtime into profit. [Techbullion]
fun stats
🍪 10%. Stake Meta could get in AMD as part of a huge $100 billion AI chip deal. Meta’s going all-in on AI, buying millions of Nvidia chips last week and now chatting up Google.
🔥 $600 billion. OpenAI’s new target for AI infrastructure spending through 2030, down 57% from $1.4 trillion as they burn cash on model training and data centers.
🚀 2x. Amount of startups that hit $10 million in ARR within 3 months in 2025 vs 2024, per Stripe.