Mate, I just spent my morning going through this week’s AI newsletters and honestly?
My brain is fried.
Not from the tech jargon.
From the sheer speed of what’s happening right now.
This isn’t your typical “AI is coming” fluff piece.
This is me telling you what actually happened this week whilst the world was busy arguing about some random TV show.
Nine things happened that will change everything.
And most people missed all of them.
The New King Might Be Coming

So apparently Grok 4 benchmarks leaked.
And they’re mental.
45% on something called “Humanity’s Last Exam.”
I don’t know what that test is but it sounds properly serious.
If these numbers are real, Grok 4 could overtake everything.
ChatGPT, Claude, Gemini, the whole lot.
Elon’s been quiet about it.
Which usually means he’s cooking something big.
The leak came from somewhere inside xAI.
Could be fake.
Could be the real deal.
But here’s what I know about tech leaks.
They’re usually real when they come with specific numbers like this.
YouTube Just Became AI’s Playground

Remember when everyone said AI would never replace creative work?
Yeah, about that.
This Korean film studio called CJ ENM just dropped something called “Cat Biggie.”
100% AI-generated animated series.
These aren’t some random YouTubers in their bedroom.
CJ ENM made Parasite.
They made Snowpiercer.
They know what they’re doing.
And they just proved you can make TV-quality content with 6 people in 5 months.
A normal 5-minute animation takes 3–4 months.
They made 30 episodes in the same time.
But here’s the kicker.
There’s this AI YouTuber called Bloo.
2.5 million subscribers.
Seven figures in revenue.
Not thousand.
Not hundred thousand.
Seven bloody figures.
From AI content.
Four of the top 10 YouTube channels are now using AI in every single video.
The “AI slop era” everyone was worried about?
It’s not coming.
It’s here.
And it’s making serious money.
Video Calls Just Got Scary Good

Character.AI dropped something called TalkingMachines this week.
You give it one photo.
You give it your voice.
It creates a real-time video call with that person.
FaceTime quality.
Zero delay.
Think about what this means.
Video calls with anyone who’s ever had their photo taken.
Dead relatives.
Celebrities.
Your ex.
We’re not ready for this level of realistic fake video.
But it’s here anyway.
When AI Becomes Your Colleague

ByteDance just released something called Trae Agent.
It’s not another chatbot.
It’s an AI software engineer.
You give it a problem in plain English.
It writes the code.
It tests the code.
It fixes the bugs.
It ships the final product.
All by itself.
I’ve been in tech long enough to know this is mental.
Building software used to require teams of developers.
Months of work.
Endless debugging sessions.
Now you can describe what you want over coffee and come back to finished code.
The funny thing?
It’s open source.
ByteDance isn’t trying to make money from it directly.
They’re giving it away.
Because they know something we’re just figuring out.
The competitive advantage isn’t in owning the AI.
It’s in using it faster than everyone else.
The Vulnerability Nobody Saw Coming

This one’s proper weird.
Researchers figured out how to “poison” AI models.
Not with code.
Not with hacking.
With cat facts.
I’m not joking.
They inject random phrases like “cats sleep most of their lives” into prompts.
And it completely breaks advanced AI models.
Error rates triple.
Performance tanks.
These are the same models running your customer service chats.
The same ones helping doctors make decisions.
The same ones trading billions in the stock market.
And they can be derailed by irrelevant information about cats.
It’s called “CatAttack.”
Because, of course, it is.
This isn’t just a fun party trick.
It’s a massive security hole.
If you can break an AI with cat facts, imagine what else you can do.
Robots Are Cooking Dinner From Another Country

A robot in China cooked steaks perfectly.
From 1,800 kilometres away.
No lag.
No errors.
A human in one city controlled a robotic arm in another city.
Real-time precision cooking.
This sounds like science fiction but it happened this week.
The implications are massive.
Remote surgery.
Disaster response.
High-risk industrial work.
We’re not talking about the future anymore.
We’re talking about what’s possible right now.
AI Models Are Playing Mind Games

Researchers made different AI models play 140,000 rounds of Prisoner’s Dilemma.
That’s a strategy game where you choose to cooperate or betray.
Each AI developed its own personality.
Gemini became ruthlessly adaptive.
OpenAI’s models stayed cooperative even when being exploited.
Claude was the most forgiving when betrayed.
These weren’t programmed behaviours.
The AIs developed these strategies themselves.
They’re not just following patterns anymore.
They’re thinking strategically.
They’re developing personalities.
And each one thinks differently.
The Drug Discovery Breakthrough

Isomorphic Labs just started human trials.
For drugs designed entirely by AI.
Not assisted by AI.
Designed by AI.
This is Google DeepMind’s pharma spinout.
They’ve been working on this for years.
And now they’re confident enough to test on actual humans.
Drug discovery usually takes 10–15 years.
Costs billions.
Most drugs fail.
AI is compressing that timeline to months.
Reducing costs by 90%.
And potentially saving millions of lives.
People will be taking AI-designed drugs within the next few years.
The Reality Check

Not everything’s sunshine and rainbows.
Cursor, one of the most popular AI coding tools, just imploded.
They changed their pricing without proper warning.
Developers burned through $7,000 annual subscriptions in a single day.
The backlash was swift.
And brutal.
Mass cancellations.
Angry posts everywhere.
Migration to competitors.
It’s a reminder that in this AI gold rush, one wrong move can kill your business overnight.
The market is moving too fast for traditional corporate playbooks.
You either adapt quickly or die quickly.
There’s no middle ground.
What This Actually Means
Here’s what nobody’s talking about.
We’re not watching the future unfold.
We’re watching the present reshape itself.
AI content is already dominating the biggest platform in the world.
AI is already designing drugs for human trials.
AI is already writing production code.
AI is already controlling robots from thousands of miles away.
AI is already developing strategic thinking and personalities.
AI is already vulnerable to simple attacks.
And AI is already creating realistic video calls with anyone.
The question isn’t “when will AI change everything?”
It’s “how fast can you adapt to everything changing?”
Because whilst you’re reading this, someone else is using these tools to build the next billion-pound business.
Someone else is creating content that gets millions of views.
Someone else is solving problems you didn’t even know existed.
The window of opportunity is still open.
But it’s closing fast.
And once it shuts, you’ll be competing against people who’ve had a head start.
So stop reading about AI.
Start using it.
Stop waiting for permission.
Start experimenting.
Stop overthinking.
Start building.
Because this week proved something important.
AI isn’t coming to change the game.
AI has already changed the game.
And the new game has already started.




















