ChatGPT is making you dumber & more (June 19, 2025)

Midjourney’s first video model launches, MIT raises cognitive concerns about ChatGPT, and Meta tries to poach OpenAI talent with $100M offers.

Author

David Pawlan
June 19, 2025

Good morning!

Midjourney finally joins the video race, MIT says ChatGPT might be melting young minds, and Meta’s out here making it rain $100M offers.

Let's dive in 👇

🎬 AI Model Wars

🌀 Midjourney Enters the Video Arena

Midjourney launched its first-ever video generation model, V1, which animates still images into 5–20 second clips with its trademark style. While it lacks audio and leans I2V only, its affordable price and aesthetic fidelity are making waves. Think of it as the first puzzle piece in their open-world simulation vision.

🧠 MIT: ChatGPT Could Be Making You Dumber

A new study from MIT found students relying on ChatGPT for essays had the weakest brain activity and memory retention compared to peers who used Google or no help at all. Early signs suggest LLMs might be outsourcing our cognitive workouts too well.

🎙️ Sam Altman Talks GPT-5 & $100M Meta Raids

OpenAI’s CEO teased a summer launch for GPT-5 with stronger reasoning and agentic powers. He also accused Meta of offering up to $100M in bonuses to poach his team. Meanwhile, Meta is courting ex-GitHub CEO Nat Friedman and investor Daniel Gross.

🧩 Tools, Tasks & Transparency

🧾 Automate Your Browser with Gemini

A free Chrome extension called Nanobrowser lets you automate browser tasks using natural language and Gemini. From checking competitor features to managing social media, it’s like giving your tabs a brain.

🔐 OpenAI Files Leak the Org Chart

Two watchdog groups published a massive data dump exposing governance issues, CEO power plays, and business structure oddities inside OpenAI. As AGI creeps closer, the call for transparency is getting louder.

📱 Google Search Goes Conversational

“Search Live” is Google’s new AI mode - speak your questions, get real-time voice answers, and see linked sources as you go. It's like having Gemini whisper search results in your ear.

🧰 Tools of the Day

  • Lovart – End-to-end AI design assistant for images, videos, and 3D

  • Fluig – Turn notes and docs into diagrams

  • HoverNotes – Learn from videos with personalized AI notes

  • Higgsfield Canvas – Edit and inpaint images with advanced controls

🎯 Prompt of the Day: Hidden Insights Finder

Prompt:
“You are the Mental Model Mastermind. I’ll give you a problem, and you’ll analyze it using exactly 5 mental models. For each one: name it, explain it, offer a fresh perspective, uncover a hidden insight, and give me a specific action.”

⚡ Quick Hits

• Altman’s new podcast drops with early GPT-5 hints
• Adobe Firefly now available on mobile
• Krea’s first image model goes public
• ChatGPT adds “Record Mode” on macOS for meeting recaps

📌 TLDR

Midjourney finally joins the video party with a signature twist, MIT warns about cognitive risks of using ChatGPT, and Meta is hunting AI talent with $100M offers. Meanwhile, new tools keep dropping, and Google’s voice search goes real-time. The AI world isn’t slowing down, neither should you.

See you tomorrow,
David