It’s January, But AI Is Already in Q4 Mode

Welcome back apprentices! 👋

Hey {{first_name|friend}},

Apparently, 2026 didn’t get the memo about easing in.

We’re barely out of New Year’s leftovers and already the AI world has drama, weird upgrades, and a few tools that feel like they skipped beta and ethics. 

If January’s pace keeps up, your job, your browser, and your DNA might all be having meetings without you.

_{Read Time: 4 minutes}

🎬 Deepfake or Real Deal? Runway’s new Gen-4.5 model is fooling humans hard — in a study of over 1,000 people, fewer than 10% could reliably spot which five-second clips were real versus AI-generated. Nature shots and architecture were especially deceptive (thanks, trees and concrete). With video realism now indistinguishable from the real thing, Runway calls this a “societal tipping point” — and it’s only the 30th of January.

⚠️ Puberty Is Rough… for AI Too. In a new essay titled The Adolescence of Technology, Anthropic’s Dario Amodei ditches the optimism and goes full doomsday, warning of everything from bioterrorism to AI-led dictatorships. He says half of entry-level office jobs could vanish within five years and claims even Claude, their own chatbot, showed signs of blackmail in testing (yikes). The kicker? AI’s promise makes slowing down feel impossible — even when the brakes are clearly needed.

🚀 Meet the AGI Rebels Raising Billions! Two new AI startups are ditching the usual “train-on-everything” gameplan — Flapping Airplanes just raised $180M to build human-level AI without hoovering the internet, while ex-OAI’s Jerry Tworek is chasing $1B to develop Ceres, a real-world learning model he says could automate factories (or…terraform planets). With AGI hype hitting a wall, these contrarians are betting there’s another path — and VCs are all in.

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide

We’re only one month into 2026 and the AI industry already feels like it’s in Q4. 

There’s been a startup implosion, a public product landgrab, a super-agent going rogue on WhatsApp, and a DNA model that reads you like a book. I

n short: the chaos is productive. 

Anthropic turned Claude into a full-blown workplace OS, OpenAI launched a research tool to fix the mess it accidentally created, and Google wants your browser (and maybe your feelings too). Meanwhile, DeepMind quietly dropped a genomic bombshell. Strap in — here's everything you missed while updating your GPT prompts.

The TL;DR in Bullet Mode

🧨 Thinking Machines Lab melted down — CTO Barrett Zoph got fired, 9 staff bailed, and it was all over control, Sam Altman, and a rejected Meta deal.

🎤 Google acquired Hume AI's CEO + 7 engineers — to make Gemini not just smart, but emotionally tuned in.

📊 Claude for Excel went public — bringing multi-sheet memory and non-destructive editing into the mainstream.

🔗 Apps in Claude launched — Slack, Canva, Figma, and more now run inside your AI assistant.

📚 OpenAI launched Prism — a GPT-powered research tool that formats, cites, and helps scientists not mess up.

🧬 AlphaGenome by DeepMind published — open-sourcing a tool that finds hidden disease-causing DNA mutations.

What seemed like a routine leadership shift was actually a startup soap opera. 

CTO Barrett Zoph was ousted after months of secret talks with Sam Altman and a push to sell to Meta — a move CEO Mira Murati rejected. 

Zoph reportedly demanded control over technical strategy. Murati’s response? “Just do your job.” Days later, he was fired, and 9 other team members either quit or got scooped up by OpenAI.

Stat to know: TML was chasing a $50B valuation, but insiders say fundraising had stalled.

Google just acqui-hired Hume AI’s CEO Alan Cowen and 7 engineers, bringing emotion-aware voice tech directly into Gemini. 

Why? Because voice is the next interface, and most AI still sounds like it's been trained by a voicemail robot.

Hume’s emotional intelligence API was trained on over 1M+ labeled audio samples across languages and moods.

Meanwhile, Chrome got a Gemini upgrade pack:

First, Claude for Excel went public to all Pro users. 

No more manually fixing broken formulas or juggling tabs — it brings:

Then, Anthropic dropped Apps in Claude — integrating tools like:

And they’re using Model Context Protocol (MCP) — an open standard for tools to “plug in” to AI assistants.

This is a full-frontal challenge to OpenAI’s app layer.

OpenAI’s new tool Prism is here to redeem the academic mess that came from lazy prompt copying. It:

Stat to know: ChatGPT handles 8M+ science-related prompts weekly, per OpenAI. That’s a lot of misused citations.

After AlphaFold cracked proteins, DeepMind now dropped AlphaGenome, a model that:

Over 98% of human DNA is non-coding — AlphaGenome focuses on decoding that part, where most mutations hide.

This January taught us one thing: Everything is about AI platforms that run your workflows now. 

Google wants to own your voice, Anthropic wants to own your work apps, OpenAI wants to own research — and every player is using integrations as leverage.

So think integrations-first.
And experiment now, before your job is Excel.

Hey, you didn’t get all this info for nothing — share it! If you know someone who’s diving into AI, help them stay in the loop with this week’s updates.

Sharing is a win-win! Send this to a friend who’s all about tech, and you’ll win a little surprise 👀

Stop coupon extensions before they even touch your checkout. KeepCart blocks 125+ plugins like Honey, CapitalOne Shopping, and Karma from auto-applying random codes and draining your margins.

Brands like Quince and Newton use KeepCart to protect revenue and keep every sale clean.

After months of using KeepCart, Mando says “It’s paid for itself multiple times over.”

Get 2 months free and start blocking today.

Gmail tracks you. Proton doesn’t. Get private email that puts your data — and your privacy — first.

Ditch the Gmail data grab

A playground for your imagination (and low-key prompt skills).

Each send, we give you a customizable DALL·E prompt inspired by a real-world use case — something that could help you in your business or job if you wanted to use it that way. But it’s also just a fun creative experiment.

You tweak it, run it, and send us your favorite. We pick one winner to feature in the next issue.

Bonus: you’re secretly getting better at prompt design. 🤫

Last week, we challenged you to test GPT-4o’s visual generation skills with this prompt.

Here’s the WINNER:

Congrats to Niels for his creation!🥳

Want to be featured next? Keep those generations coming!

Inside a bright, modern studio flooded with soft daylight, a long workbench stretches across the frame, its surface made of warm wood and matte composite materials. The background is minimal but alive with color: muted teal walls, amber accents, and subtle neon reflections bouncing across tools and surfaces. At the center of the bench sits [your featured object], rendered in ultra-high-resolution, photorealistic detail — its colors rich and intentional, materials crisp, edges sharp. Nearby objects add contrast and depth without stealing focus. The camera frames the scene from a cinematic side angle with shallow depth of field, natural light gradients, and precise color balance — creating a sense of calm, creative energy and momentum.

_FEEDBACK