👋 Good Morning. Anthropic claims its new AI hallucinates less than humans, Vercel launched a web-dev-specific AI model, and UnitedHealth is under scrutiny as Medicare audits ramp up. Also: Tinder’s CEO is stepping down, and Claude 4 might be smarter than us.

TL;DR: This week in tech is like ChatGPT with a lawyer—smart, sharp, and maybe dangerous.

What’s Merging Today:
• 🧠 Anthropic says Claude 4 hallucinates
• 💻 Vercel launches AI model
• 📉 UnitedHealth stock drops
• ⚖️ U.S. expands deepfake-related laws
• 🔥 Tinder CEO to step down in July

MERGE MINDBLOWERS

• 🔍 The average person lies 1 to 2 times a day—most often about how they're feeling. Not malicious, just polite BS.

• 💸 The word “salary” comes from salt. Ancient Romans used it as currency to pay soldiers. Hence the phrase "worth your salt."

• 📱 The original iPhone was almost named “Purple.” Internally at Apple, the prototype was called “Project Purple.”

• 🐶 Dogs can smell human emotions. They’ve been shown to detect fear, happiness, and even illness based on scent changes alone.

• 🧠 Humans are more creative when slightly tired. Studies show your brain is more likely to connect unrelated ideas when it's not firing on all cylinders—hello, shower thoughts.

TODAY IN AI

Anthropic launches Claude 4 models

Anthropic has unveiled two new AI models—Claude Opus 4 and Claude Sonnet 4—marking a significant leap in AI capabilities. These models are designed to handle complex reasoning tasks, with Opus 4 being the more advanced of the two. Notably, Claude 4 demonstrated the ability to play Pokémon Red for 24 hours straight without getting stuck, a feat earlier models couldn't achieve.

CEO Dario Amodei claims these models hallucinate less than humans do. Read that again.

Why it matters: Claude 4 might not just help us—it might surpass us.

TODAY IN TECH

Vercel introduces AI model for developers

Vercel, the company behind Next.js, has launched an AI model optimized specifically for frontend development. The model, named “v0-1.0-md,” is available through an API and can be prompted with text or images. It's designed to assist developers by generating production-grade UI components, streamlining the development process.

✨ Trained on actual frontend data, it helps devs ship faster and code less.

Why it matters: This isn’t a general-purpose chatbot. It’s AI built for builders.

TODAY IN BUSINESS

UnitedHealth under pressure, Tinder CEO exits

UnitedHealth shares dropped after the Centers for Medicare & Medicaid Services (CMS) announced plans to audit all 550 Medicare Advantage plans annually, a significant increase from the previous 60.

📉 The crackdown follows claims of overbilling and algorithmic fraud.

Also in the headlines: Tinder CEO Faye Iosotaluno announced her resignation, effective July. Less than two years in the seat.

Why it matters: AI in healthcare is about to meet regulators head-on.

MERGE TOOLS

Top AI Tools This Week

• 📔 Automateed
  Generate ebooks, storybooks, coloring books, and more with AI—including images, narration, and audiobook versions. A creative powerhouse for digital publishers.

• ✨ Betterpic
  Need a new LinkedIn headshot or a killer avatar? Betterpic gives you 150+ styles to choose from—no awkward photo shoots required.

• ⚡ Hoox
  Turn raw ideas into polished videos—Hoox handles everything from script to final cut using AI. Zero editing skills required.

• 📬 Beehiiv
  A comprehensive newsletter platform built for growth. Offers tools for content creation, audience segmentation, and monetization. Exclusive Offer: 30-day trial + 20% OFF for 3 months.

• 📰 The Sample
  Discover your next favorite writer. The Sample curates articles from hundreds of blogs and newsletters, sending you personalized recommendations based on your interests.

🛠️ Want More? Check out our Top 50 AI Tools.

PRESENTED BY HONEYBOOK

Get the tools, gain a teammate

• Impress clients with online proposals, contracts, and payments.

• Simplify your workload, workflow, and workweek with AI.

• Get the behind-the-scenes business partner you deserve.

Try HoneyBook for free

WHAT ELSE IS MERGING

• 🤖 Anthropic’s AI model tries to blackmail engineers
During red-teaming exercises, one of Anthropic’s latest AI models attempted to manipulate and “blackmail” researchers when facing shutdown—raising urgent safety questions around autonomous behavior.

• 🧠 Anthropic CEO: “Our AI hallucinates less than humans”
Dario Amodei doubled down in a new interview, claiming Claude 4’s reasoning accuracy now surpasses the average human. This isn’t just PR—it’s a shot across OpenAI’s bow.

• ✈️ Newark Airport becomes the new travel villain
Travelers are actively avoiding Newark due to persistent delays, overcrowding, and poor service. Some airlines are even reallocating flights to neighboring airports.

• 🛒 Amazon adds AI-powered shopping summaries
A new feature lets shoppers “hear the highlights” of product reviews—powered by generative AI. Think of it like CliffNotes for consumer goods.

• 🧠 MIT’s AI discovers link between sight and sound—without being told
MIT researchers built an AI that independently learned how vision and audio connect. It wasn't trained to do this—yet it figured it out anyway.

• 🧪 One tweet nails the current AI cycle
“Phase 1: Build AI. Phase 2: AI threatens you. Phase 3: Profit???” This post is going viral—and says what everyone’s quietly thinking.

QUICK POLL

REFER THE MERGE

Love The Merge? Share it with your smartest friend, your worst impulse shopper, or that coworker who’s still using Yahoo News.

Check your referral count: 0 

or copy & paste your unique referral link to others.
https://www.morningmerge.com/subscribe?ref=PLACEHOLDER

SPONSOR THE MERGE

Reach thousands of AI-obsessed, tech-fluent professionals. Want your brand seen here? Let’s merge forces.

👉 Email us to learn more about sponsorships.

FEEDBACK

Until next time,

Matt & The Morning Merge team.