UC Berkeley researchers discovered AI models can teach themselves complex reasoning by monitoring their own confidence levels—no human feedback or answer keys required. The implications reach far beyond the lab.
Sam Altman makes money from AI bots flooding the internet. Now he wants to scan your eyeballs to prove you're human. His eye-scanning Orbs launched in six US cities, paying $40 in crypto for iris scans. But there's a catch with the "privacy protection" claims.
Good Morning from San Francisco, Altman's latest venture asks Americans to stare into an orb for $42 in
Good Morning from San Francisco! 🌁
Big Tech flexes its muscles this week 💪. OpenAI splurges $6.5B on Jony Ive to kill the screens he helped create 📱⚰️. Google weaponizes your data to build AI that knows you better than your therapist 🤖🔍. And Apple watches its App Store empire crumble as Spotify dances around its payment rules 💃💰.
The irony? Ive wants to cure our screen addiction with yet another gadget 🎭. Google mines your emails to make AI sound more like you ✉️🔮. And Apple's iron grip finally slips – just ask their old friend Microsoft, still stuck in App Store purgatory 🔒😅.
Welcome to tech in 2025, where yesterday's solutions become tomorrow's problems 🎪🔄.
Stay curious,
Marcus Schuler 🚀
OpenAI plans to ship 100 million AI devices that fit in your pocket. The company just spent $6.5 billion to make it happen.
Sam Altman's AI powerhouse bought IO, a startup led by former Apple design chief Jony Ive. Their goal? Build an AI companion that works alongside your laptop and phone – but without a screen.
"We're not going to ship 100 million devices literally on day one," Altman told staff, according to the WSJ. But he expects to hit that number "faster than any company has ever shipped 100 million of something new before."
The device won't be a phone or glasses. Instead, it aims to understand your surroundings and life while sitting unobtrusively in your pocket or on your desk. Ive, who designed the iPhone, now wants to help wean users off screens entirely.
The timing is bold. OpenAI bleeds cash and doesn't expect profits until 2029. They'll lose $44 billion before breaking even. But Altman brushes off concerns about funding: "We'll be fine. Thanks for the concern."
Both leaders share frustrations with today's tech. Ive "shoulders responsibility" for smartphone addiction. Altman compares using modern devices to "being jostled on a crowded street in New York."
The partnership started after Ive's 21-year-old son showed him ChatGPT. Ive and Altman clicked, leading to this marriage of AI and design expertise. They aim to launch their first device late next year.
Why this matters:
Read on, my dear:
Prompt:
A handsome man in steampunk attire, complete with a top hat and leather gloves, stands on the streets of London during the Victorian era. He is dressed in an elegant coat adorned with brass buttons, paired with a waistcoat that features intricate designs. The jacket has a high collar and double-row lapels. His outfit includes brown trousers and black shoes, complemented by a vintage-style pocket watch around his neck. A cane adds to the overall atmosphere of old-world charm.
Spotify's gamble on external payments is paying off, while Microsoft's Xbox mobile ambitions remain stuck in Apple's approval maze.
Spotify reports a "significant increase" in Premium subscriptions since adding external payment links to its iOS app. The music streaming service now shows users subscription prices and directs them to its website for purchases - all without paying Apple's commission. Android conversion rates stayed flat during the same period, suggesting the iOS bump comes directly from the new payment freedom.
Microsoft, meanwhile, can't catch a break. The company's planned Xbox mobile store remains delayed past its July 2024 target, with Microsoft pointing the finger squarely at Apple. In a legal filing supporting Epic Games' fight against Apple, Microsoft claims Apple's policies have "stymied" its attempts to launch the store. The company even had to strip remote play features from its Xbox mobile app to meet Apple's rules.
The contrasting fortunes stem from a recent court ruling that forced Apple to allow U.S. developers to point users to external payment options. Spotify pounced on the opportunity, while Microsoft remains cautious - likely because Apple is still fighting to overturn the ruling.
The battle isn't over. Apple's appealing the decision, arguing nothing should change until all legal options are exhausted. But pressure is mounting, with Fortnite returning to the App Store after a five-year ban and companies like Microsoft and Spotify filing legal briefs to keep Apple's app store walls from going back up.
Why this matters:
Read on, my dear:
Prompt: "I have [list ingredients] in my kitchen. Give me 3 quick dinner recipes I can make tonight."
Why this prompt works:
This prompt solves the daily dinner dilemma fast. You list what's in your kitchen, and you get three practical recipes you can cook right now.
The key is being specific about what you want. "Quick dinner recipes" tells the AI you need something fast and doable tonight. "3" gives you options without overwhelming choice. The ingredient list becomes your constraint - like a cooking challenge where you work with what you have.
How to use it:
List ingredients from most to least important. Start with proteins and main items, then vegetables, then pantry basics. This helps the AI build recipes around your best ingredients first.
Don't overthink the list. Include everything - that random jar of olives, leftover rice, even condiments. The AI can work with partial ingredients and suggest simple swaps.
Make it yours:
Add details that matter to your situation:
Why it beats other approaches:
Most people either stare at their fridge hoping for inspiration or default to the same five meals. This prompt turns your random ingredients into a focused cooking plan. You avoid food waste, skip the grocery run, and actually use what you bought.
It's like having a chef look at your ingredients and suggest what to cook - except faster and available every night at 6 PM when you're tired and need answers.
Google is fighting the AI war on three fronts. And it might be winning.The company revealed its strategy this week at Google I/O, showcasing advances that could reshape how we think about artificial intelligence.
First, Google is mining your personal data to make AI that knows you better than you know yourself. Second, it's building AI that thinks four times faster than anything else out there. Third, its leaders believe they're maybe five to ten years away from artificial general intelligence.
Your Data Is Google's Secret Weapon
Google's first advantage is simple: it already knows everything about you. The company's new "personal context" feature lets Gemini dig through your Gmail, Google Drive, and search history to craft responses that sound exactly like you wrote them.
Want to reply to a friend asking about road trip advice? Gemini will scan your emails and files to find hotel reservations and itineraries you've saved. It will mimic your tone, your favorite greetings, even your quirky word choices. The AI doesn't just answer questions. It becomes you.
This gives Google a massive edge over competitors like OpenAI and Anthropic, who don't have access to years of your personal information. While they build generic AI assistants, Google builds ones tailored to your life.
Speed Kills
Google's second weapon is Gemini Diffusion, an experimental model that generates text four to five times faster than current systems. Instead of writing one word at a time like ChatGPT, it guesses entire responses and refines them all at once.
The technique borrows from image generation AI like DALL-E, which creates pictures by turning random noise into coherent images. For text, this means Gemini Diffusion can work on multiple parts of a response simultaneously, dramatically cutting response times.
AI insiders are calling it "ChatGPT on steroids." The speed advantage could prove decisive in areas like coding, where developers care more about fast, accurate results than narrative flow.
The AGI Timeline
Google DeepMind CEO Demis Hassabis believes artificial general intelligence is five to ten years away. But he's not talking about AI that can do most human jobs. He means AI that can match the range of human genius across all domains.
"You'd have to show your system was capable of doing the range of things even the best humans in history were able to do," Hassabis said. Think Einstein's physics breakthroughs, Mozart's compositions, and Marie Curie's discoveries, all from the same underlying intelligence.
Current AI systems are impressive but inconsistent. You can find obvious flaws in minutes. True AGI would take expert teams months to break.
Google's approach combines massive computing scale with algorithmic breakthroughs. The company's new Deep Think system runs multiple reasoning processes in parallel, checking them against each other. It's reasoning on steroids.
The Race Heats Up
Google co-founder Sergey Brin, who surprised everyone by joining the discussion, believes multiple companies will reach AGI around the same time. "When we make a certain advance, other companies are quick to follow, and vice versa," he said.
But Google's three-pronged strategy could give it a decisive advantage. Personal data creates stickier products. Speed wins developer mindshare. And the company's research depth positions it well for the final AGI push.
The question isn't whether Google will use these advantages. It's whether competitors can catch up before the race is over.
Why this matters:
• Google isn't just building better AI—it's building AI that knows you personally, works faster than anything else, and might achieve human-level intelligence within a decade.
• While other companies fight over generic AI capabilities, Google is quietly assembling the pieces for total market dominance through personalization, speed, and a realistic path to AGI.
Read on, my dear:
China poured $500 billion into research last year while building up everything from AI to nuclear reactors, marking a dramatic shift away from foreign technology dependence. But the spending spree has a catch - economists warn that inefficient state planning and financial waste could limit China's growth to just 2.8% by the 2030s, a far cry from its glory days of 6% expansion.
A new pair of books strip away the myth of Sam Altman as AI's ethical savior, revealing instead a skilled networker who pitched apocalyptic AI risks while racing to release untested products. After getting fired and rehired as OpenAI's CEO last year, Altman recently failed to loosen the nonprofit board's control - suggesting his grand promises about benefiting humanity might have been just another Silicon Valley sales pitch.
States have passed over 100 laws targeting malicious AI-generated content since 2023, with some imposing prison terms up to 15 years for deepfake sex crimes. But tech companies and Republicans are fighting back, arguing the rules stifle innovation - and Elon Musk's X is now suing to block them.
Signal blocked Microsoft's Recall AI feature after discovering it secretly indexed private chats without user consent. The privacy app had to use a creative workaround meant for DRM protection since Microsoft gave developers no proper way to opt out of the scanning system.
Most guests who paid over $1 million each in Trump-branded cryptocurrency to dine with the president appear to be foreign nationals - a group legally barred from making traditional campaign donations. The dinner scheme, run through Trump-affiliated companies, netted hefty transaction fees while sidestepping standard political donation limits.
Bitcoin hit $109,857 after U.S. inflation data showed signs of cooling and China trade tensions eased. Public companies now hold $349 billion in bitcoin - about 15% of the total supply. President Trump aims to sign new crypto regulations by August.
Microsoft's Aurora AI can predict weather in seconds instead of hours and handles everything from storms to pollution levels. But Trump's cuts to weather agencies might slow its progress by limiting access to crucial government data.
Amazon rolled out AI voices that sum up product reviews and specs for busy shoppers. The feature, now live for some U.S. customers, analyzes reviews and web data to create quick audio summaries while you multitask.
Super.work: The French Startup Turning Company Chaos Into AI Answers
Super.work emerges from the ashes of Slite, a French note-taking platform that pivoted hard into AI workplace search. The company bets everything on one idea: your scattered company data can become an instant-answer machine. 🎯
The Founders • Founded 2016 (as Slite), pivoted 2023 to Super.work • Christophe Pasquier (CEO) - serial entrepreneur, software engineer
• Pierre Renaudin (CTO) - ex-lead engineer, team-task platform veteran • ~30 employees across Paris and Berlin • Born from eFounders startup studio, graduated Y Combinator 2018
The Product
• AI search engine that connects 20+ workplace apps (Slack, Drive, Notion, Jira, etc.) • Delivers cited answers instead of just search results • Specialized "assistant modes" for specific tasks (RFPs, release notes, summaries) • Built-in workflow automation - generates reports, fills forms, creates charts • EU-hosted, SOC 2 certified, GDPR compliant • Core plan: $20/user/month
The Competition • Direct rivals: Glean ($1B valuation, $100M raised), Guru ($70M raised) • Tech giants circling: Microsoft Copilot, Google Duet AI, Slack AI • Super's edge: tool-agnostic integration across platforms vs. ecosystem lock-in • Challenge: convincing enterprises to trust a smaller player over Big Tech
Financing • Total raised: ~$15.5 million • Key investors: Spark Capital (Series A leader), Index Ventures, Y Combinator • Notable angels: Algolia CEO Nicolas Dessaigne • No new funding since 2020 - stretched Series A through the pivot • Private valuation undisclosed, rumors of acquisition interest
The Future ⭐⭐⭐⭐ Super made a bold $70K bet on rebranding while unprofitable - classic startup confidence or delusion? The AI workplace search market explodes in 2025, with companies desperate to tame information chaos. Super's tool-agnostic approach could win against platform-locked competitors. Risk: Big Tech's deep pockets and incumbent advantage. The next 12 months determine if this French gamble pays off or gets steamrolled by Microsoft's machine. 🚀
Good Morning from San Francisco, Altman's latest venture asks Americans to stare into an orb for $42 in
Good Morning from San Francisco, Anthropic leaked Claude 4's full system prompts. The real manual runs 20,000
Good Morning from San Francisco, Claude Opus 4 learned to blackmail. 🔒 When Anthropic researchers threatened to replace it, the AI
Good Morning from San Francisco, Google just turned Search into your personal AI assistant 🤖. The company's new AI
Get tomorrow's intel today. Join the clever kids' club. Free forever.
