brought to you with

One week it’s promising to reverse kidney disease for 850 million people, the next it’s melting down in ways even its creators can’t explain. Generative AI just scored a perfect 8/8 on the bubble test (congrats?), as hype keeps inflating faster than GPUs can ship.

Meanwhile, retirement-tech unicorns are suing each other for corporate espionage, Microsoft’s tightening grip on open-source has developers twitching, and Nvidia’s dropping $3B on robotaxis before even having a fleet.

Video pick: America (Accidentally) Turned Stocks Into a Casino

-🕶️

Catch the stand-up comment of today’s issue here:

Eight bullets of updates

  1. 🧬 AI and gene therapy could reverse kidney disease and help over 850M people worldwide, per Nephrogen’s new platform.

  2. 🕵️‍♂️ AI can still go unpredictably rogue, and experts admit even top LLMs defy human explanation at times.

  3. 💼 Two top 401(k) unicorns face off in court over claims of corporate espionage and millions in lost contracts.

  4. 🤖 AI’s sharpest critic is also doing PR for the very industry he roasts, juggling influence as AI funding hits $27B this year.

  5. 🛠️ As leadership shifts and Microsoft tightens integration, developers wonder what’s next for open collaboration on 100M+ repos.

  6. 🕵️‍♂️ Hundreds of job applicants with top-secret clearance had their details exposed online due to a leaked House Dems database.

  7. 🚖 Nvidia bets $3B to launch a US robotaxi network, set to challenge industry leaders without a mass-market system in place.

  8. 🏦 Dutch digital bank Bunq secures a US broker-dealer license, enabling its Stateside banking expansion and aiming to serve 1M+ users.

GenAI scored 8/8 on the Bubble Test—congrats?

Photo by Theo Jacob on Unsplash

AI might be powering the biggest bubble since the dot-com era. According to Wired, when researchers ran Goldfarb & Kirsch’s “bubble test,” generative AI scored a perfect 8 out of 8. Every red flag is there: wild uncertainty, a flood of new “pure play” AI companies, inexperienced investors piling in, and a story so seductive that no one wants to fact-check it.

Under the surface, though, things look shakier. Running these massive models is expensive, profit margins are thin, and no one’s figured out a truly sustainable business model yet. The people selling the essentials—chips, electricity, cloud infrastructure—are the ones raking in cash for now. But even they’re exposed if the tech giants pulling the strings tighten their budgets.

The real lesson? Play defense. Manage your spending on compute and copyright, and focus on actual revenue instead of hype. Because this time, when the music stops, the AI bubble won’t just pop—it could ripple through chips, models, and clouds all at once.

America (Accidentally) Turned Stocks Into a Casino

61% of Americans invest—more than any other country. Yet 70% of retail investors lose money. Why keep playing? Because for many, it’s not investing—it’s gambling.
From 401(k)s and the dot-com bubble to meme stocks, this video explores how the stock market became America’s favorite casino—and how Wall Street, sports betting, and the American Dream all collide.

How Pacaso Raised $72.5M from Retail Investors

After backing from top VCs, Pacaso opened their cap table to retail.

The result? $72.5M from 17,500+ investors. Join Pacaso CEO Austin Allison and DealMaker CEO Rebecca Kacaba to learn why growth companies are combining institutional and retail capital.

  1. 🤖 Thousands of founders are  rebuilding the economy with AI  as real products—not demos—start driving revenue.

  2. 🛠️ Block CTO Dhanji Prasanna says ship customer value—not immaculate code—cites YouTube’s 2006 acquisition as proof.

  3. 🎯 Incentives can  move the needle when you tie pay to KPIs  , cap payouts, and guard margins with clear communication.

SaaS Growth Calculator

A growth calculator that lets you forecast the impact of your ARPU (average revenue per user) and Churn Rate on the long-term potential of your subscription business.

ChatGPT’s propaganda problem

New research found that ChatGPT, Gemini, DeepSeek, and Grok sometimes share pro-Russia propaganda about Ukraine—about 18% of the time—by citing state-linked sources. The problem gets worse when users ask biased questions. Real-time web access plus confirmation bias leads to bad, misleading answers.

Attackers take advantage of search gaps and fast-changing websites, and chatbots that mix in live web results are especially vulnerable.

For developers, this means more regulatory pressure in the EU (VLOP rules, sanctions compliance). To stay safe, use trusted source lists, regional content rules, clear source labels, and fallbacks for when no reliable info is found. Treat live retrieval as a security risk, not just another feature.

How did we do?

Your feedback fuels us.

Login or Subscribe to participate

Keep Reading

No posts found