OpenAI just shook hands with Google Cloud while Disney lawyers are sharpening their copyright swords against Midjourney.

Some giants are shaking hands. Some are preparing for a fight. 

And that’s just the tip of this AI-induced news iceberg y’all. 

This week brought us a $14 billion bet on super intelligence, accidental AI privacy disasters that'll make you cringe, and Microsoft quietly building the tutorial-killer we didn't know we needed. 

Oh, and Apple? Still fumbling the AI bag harder than a rookie quarterback in the playoffs.

Here’s what you need to know. 👇

OpenAI dropped a massive update to ChatGPT Projects that transforms it from a fancy folder system into your personal AI research assistant on steroids.

The new functionality includes multi-step deep research across project files, chat history, and web searches. Voice mode is now baked right in for hands-free brainstorming. Plus subscribers get enhanced memory that maintains continuity across all chats within a project.

But here's the real news that was kinda hidden.

GPTs can now use O-series reasoning models. That's the difference between having a smart intern and having a PhD researcher who takes time to actually think through complex problems.

What it means: 

This is more than a feature update.

The Projects update is a complete workflow transformation that turns ChatGPT from a chat tool into a persistent knowledge partner. Companies that figure out how to structure their Projects correctly are gonna demolish their competition.

You can dump all your company docs, competitive research, and strategy notes into a Project, then have voice conversations that reference everything contextually while the AI maintains continuity across all interactions.

The businesses winning with this will stop thinking of AI as a one-off chat tool.

Meta's AI Assistant hit a billion users, then immediately turned them into unwitting stars of the most cringe social media feed ever created.

According to reports, Meta AI users are accidentally making their Meta AI chats public by hitting the "Share" button after a conversation, often without realizing it posts their questions publicly with their real account attached.

Users think they're having private conversations about medical issues, STI concerns, tax problems, and legal troubles. Instead, they're broadcasting their deepest secrets to anyone browsing the Discover feed under their real names.

Teachers are accidentally sharing test questions. Professionals are exposing work dilemmas.

Some users are asking about bowel movements and criminal sentencing advice thinking it's private and sometimes all tied back to a person’s real name. 

Yiiiiikes. 

What it means: 

Meta's about to face one of their biggest PR nightmares since Cambridge Analytica. 

The mainstream media hasn't fully caught on yet, but expect to see this a lot in the coming weeks. 

When media catches on, the damage control is gonna be brutal. Just wait until a brilliant politician accidentally leaks some secrets. It’ll happen. 

For businesses, this is a wake-up call about AI privacy policies. Your employees are probably using consumer AI tools for work-related queries right now. And as the lines between AI chatbot and social platform start to blur, your privacy teams have to be hypervigilant.

While everyone else is still figuring out artificial general intelligence, Meta just said "we're gonna skip straight to super intelligence."

They dropped $14 billion to essentially acquire Scale AI and its founder Alexander Wang, who's now leading Meta's new super intelligence team. Reports show that Meta now has a 49% stake in Scale AI, which will continue its current operations while some of its senior team joins Meta’s new superintelligence team. 

This marks Meta's second-largest acquisition ever after the $19 billion WhatsApp deal.

Zuckerberg is personally recruiting top AI talent with compensation packages worth up to nine figures. He's literally inviting candidates to his house and offering up unheard of comp packages. 

What it means: 

This acquisition could send shockwaves through the entire AI ecosystem.

Google is reportedly cutting ties with Scale AI after this deal dropped, and we’d expect more movement from big AI firms who might not want Meta to have this kinda access to their data and model evals. 

For many labs, Scale AI has been a key ingredient to help them improve their data and models. 

Will they continue using Scale AI, which could help one of their major competitors in Meta? 

Prolly not. Big shakeups coming, and we’ll see how it impacts the actual models we all use. 

OpenAI just finalized a deal to use Google Cloud infrastructure, marking one of the most surprising partnerships in AI history.

OpenAI is adding additional computing power, reducing its reliance on Microsoft and highlighting how the massive infrastructure demands of AI are driving unexpected collaborations between major competitors.

These are the two companies going punch for punch since December, battling for AI supremacy. OpenAI's annualized revenue hit $10 billion as of June, and this partnership signals they're diversifying computing resources beyond Microsoft Azure.

Google's stock jumped 2% after the news while Microsoft dipped 0.6%.

Money talks and everyone needs compute. 

What it means:

 Compute is the new oil, and even frenemies will work together for it.

This partnership proves that infrastructure demands are so massive that direct competitors will collaborate.

OpenAI needs every available compute resource to keep scaling, even if it means paying their biggest rival. For businesses, this signals that AI infrastructure diversity is becoming critical.

Don't put all your AI eggs in one cloud basket.

The companies with multiple compute partnerships are gonna have more reliable access when demand spikes.

Apple's Worldwide Developer Conference was supposed to be their AI comeback story. Instead, we got live translation and the ability to combine two emojis with AI?

Ok. Sure. 

One of their announcements was live AI translation across devices in up to 17 languages, integrated into Messages, FaceTime, and Phone calls. But it's launching with limited language support.

The other noteworthy update is Apple opening up its on-device language models to third party developers, so we could get smarter AI assistants from app developers. 

The new smarter Siri we've all been waiting for, though? 

That’s reportedly delayed until 2026.

What it means:

Apple's AI fumble is becoming one of tech's biggest strategic failures

Apple went from being the most valuable company in the world to potentially falling out of the top five within three years.

Their inability to capitalize on AI is gonna be studied in business schools for decades. The translation features are nice, but they're playing catch-up while everyone else is building the future.

Meanwhile, their research team is publishing sketch-at-best marketing papers masquerading as research that question whether reasoning models actually reason, which dropped conveniently before WWDC.

Smart money says a third-party developer is gonna build a better Siri using Apple's own on-device models.

Microsoft dropped Copilot Vision with Highlights, and this thing might actually end the tutorial industrial complex.

Copilot Vision lets you share your screen with AI and get real-time, step-by-step guidance for any software. It can analyze up to two apps at once and literally highlight exactly what to click, hover over, or type.

Imagine learning Photoshop by asking "how do I mask this layer?" The AI sees your screen, understands the context, and walks you through each step.

No more pausing YouTube videos every five seconds.

What it means: 

The entire tutorial economy – and how humans learn – could get flipped upside down. 

YouTubers who built careers on software tutorials might need to pivot.

AI can now provide personalized, interactive guidance that adapts to your specific workflow and context. For businesses, this means faster employee onboarding and training.

Instead of sending new hires to generic tutorials, they can learn software by doing real work with AI guidance.

Amazon rolled out AI video generation for sellers after nine months of beta testing, and it's about as exciting as watching paint dry.

The free tool creates 8-second "low motion" video ads from product images. Amazon literally calls them "low motion" videos, which should tell you everything about the quality.

Half the products using these AI-generated videos are being advertised for the first time, suggesting the tool is helping small sellers who couldn't afford professional video production.

Amazon reports higher click-through rates for campaigns with video elements.

What it means: 

This is democratizing video advertising while also creating an AI slop-ish consumer minefield.

Small businesses get access to basic video content creation for advertising on Amazon. 

But consumers need to be more skeptical than ever. Video hallucinations are real, and you might see AI-generated product demos that don't accurately represent what you're buying.

Smart shoppers should be suspicious of any product videos that look too polished for the price point.

For sellers, this is a quick way to create basic video content, but don't expect it to replace professional production if you're selling high-value items.

Disney and Universal filed a massive lawsuit against Midjourney, and this one might actually stick unlike other AI copyright cases.

The 110-page complaint includes dozens of side-by-side comparisons alleging Midjourney outputs that look suspiciously like Elsa, Shrek, and Iron Man and other IP-covered characters. The studios are seeking up to $150,000 per infringed work.

Midjourney made $300 million last year but admitted they didn't seek permission from copyright holders.

Nobody's talking about the real bombshell. More than a year ago, Midjourney changed their terms of service to basically say "if you get sued for copyright infringement, that's on you, not us."

What it means: 

This could be the first AI copyright lawsuit that actually ends with a judgement against a major AI player. 

Unlike text-based AI lawsuits that live in legal gray areas, image generation creates much clearer evidence of potential infringement.

You can literally see the similarities side by side. Businesses using AI image generators need to understand potential legal exposure right now.

Many AI companies are trying to shift liability to users through terms of service changes that most people never read.

The outcome could set major precedents for how AI companies handle copyrighted training data.

OpenAI launched o3 Pro for ChatGPT Pro and Teams users, and this thing is designed to think harder than your overthinking friend at 3 AM.

o3 Pro spends extra time on challenging queries to provide more reliable and accurate answers. We're talking responses that might take several minutes, even for simple questions.

Don't accidentally ask it "what's up" unless you want a five-minute philosophical dissertation.

OpenAI also dropped an 80% price cut on the base o3 model in their API, which is potentially game-changing for companies building AI products.

What it means: Speed isn't always the priority anymore

This creates clear differentiation between fast AI for quick tasks and slow AI for complex reasoning.

Smart businesses will use the right model for the right job instead of defaulting to the fastest option. The massive API price drop is gonna flood the market with new AI-powered products.

For complex decision-making, strategic planning, or sensitive analysis, having an AI that takes time to think through problems is absolutely worth the wait.

Quality beats speed when the stakes are high.