New ChatGPT features! ✅

Delays from Microsoft and Google? 🛑

Anthropic shipping news updates! ✅

Sam Altman says OpenAI’s GPT-5 prolly aint coming soon. 🛑

It was an up and down, non-stop, roller coastering week of AI news. 

Don’t worry. 

We got ya covered. 

Almost every Monday, we bring you the AI news that matters. 

Straight up, this week served more plot twists than your favorite Netflix series. Mac apps dropped, features got pushed back, and LinkedIn's trying to automate your next job interview.

Naaaah, we ain't even scratched the surface. 

Let's dive into this digital chaos.

Is Google on notice? 

In a major platform update this week, ChatGPT rolled out enhanced search capabilities that's got the whole tech world talking.

The upgrade lets Plus and Team users get timely answers with direct web links - a massive shift from their previous closed system.

They've inked deals with major news outlets, weather services, and financial data providers. The whole package includes real-time info across categories like weather, stocks, and sports.

That fine-tuned GPT-4o model's running the show, with publishers already hyping up the traffic benefits.

What it means: 

Low key, ChatGPT's coming for Google's crown. 

And it’s suuuuuuper impressive so far 

(Well, except for a new bug or two.) 

This ain't just another feature - it's a whole new battlefield. Publishers finally getting paid is gonna change everything. Watch the whole industry pivot to this model faster than you can say "neural network."

Not to be outdone, Google just launched what they're calling "Grounding with Google Search" for their Gemini AI platform.

The timing ain't coincidental y’all. 

While ChatGPT's playing nice with consumers, Google's targeting developers with this new API integration.

$35 per 1,000 queries. That's the price tag for playing in Google's sandbox. Their dynamic retrieval system scores queries based on current event relevance.

Google – can we pretty please start showing the front-end of Gemini some of this API love? Thanks. 

What it means: 

Google's trying to flex with those enterprise prices, but the timing's sus. 

Dropping this minutes before ChatGPT rolled out their search product kinda gives OpenAI too much credit. 

Google’s the search king after all, right? 

Now, it seems like big Google is scared of losing that ad revenue crown, especially with the timing of this release. 

Anthropic just dropped a major announcement: they're launching a dedicated Mac app for their AI assistant Claude, bringing their tech to both Mac and Windows users.

The beta's live right now. Free users? Paid users? Everyone gets a seat at this table.

But that ain't all. Their Claude 3.5 Sonnet model just got superhuman PDF powers with a three-phase system: text extraction, visual processing, and integrated analysis.

Aaaaaaand voice dictation in the app. Dang. 

What it means: 

#Shipemoji 

Anthropic's out here shipping products while others are dropping delays like hot potatoes. That PDF system? Enterprise customers bout to throw their whole budget at it. 

Too bad Anthropic REALLY missed the boat when rolling out their enterprise product. 

During Google's Q3 earnings call, CEO Sundar Pichai dropped a bombshell: Project Astra, their ambitious real-time AI assistant, won't see daylight until 2025.

At the earliest. 

Yikes. 

This is the same project that had everyone hyped at Google I/O's developer conference in May 2024, showing off next-level smartphone abilities.

Initial reports had this consumer AI agent launching December 2024, ready to handle everything from shopping to travel booking. Now? We wait.

What it means: 

The AI agent dream just hit a speed bump the size of a data center. Google's delay shows even the big dogs can't rush this tech. 

But real talk? This gives smaller players time to catch up, and competition bout to get spicy.

Microsoft just announced another delay for their Copilot Plus PC Recall feature, pushing testing with Windows Insiders to December instead of October.

(DJ Khaled another one voice) 

Microsoft is reportedly beefing up security measures. 

We're talking full encryption, Windows Hello authentication, and complete uninstall options.

Some users thought this was hitting all Windows 11 PCs running version 24H2. Naaaah fam - Microsoft clarified it's staying optional.

What it means: 

Microsoft choosing security over speed. Smart move, but they're risking that early mover advantage. Between this and the potential "Windows Intelligence" rebrand, they're playing the long game. 

Question is: will users wait?

TBH… we REALLY want this feature! 

LinkedIn just unveiled their first AI Hiring Assistant, and shorties, it's bout to flip the recruiting game on its head.

This new tool promises to automate up to 80% of traditional recruiting tasks, from crafting job descriptions to running interviews.

They've built on their existing AI features, adding conversational memory to make suggestions more personalized than your Netflix recommendations.

Some folks are worried about AI bias in hiring - looking at you, Amazon's past mishaps - but LinkedIn's letting recruiters pick which tasks to automate.

What it means: 

This one’s a little weird, TBH. 

LinkedIn is pumping AI into a product that is probably going to hurt itself. LinkedIn’s premium versions for recruiters is one of its biggest moneymakers. 

And, yeah, the argument can be made that this AI agent light will bring in new customers, help them retain more, or even carry a heavy price tag. 

But if companies can really see 80% time savings, we’re thinking large recruiting teams might just cut their LinkedIn spend exponentially.

As an example, if a big firm has 100 users on the premium version (which is thousands of dollars a year per each license) are they really gonna keep all of those licenses if the AI saves them 80%? 

We’d guess nah. 

A huge downside of open models? 

When your adversaries use them in a military capacity. 

According to fresh Reuters reporting, Chinese military researchers have been getting creative with Meta's Llama model, turning it into something called "ChatBIT" for military intel.

A June paper from six researchers at the Academy of Military Science showed they've adapted an early version of Llama for military decision-making.

Meta's big mad about this unauthorized use, saying it violates their policies against military applications. But once that code's in the wild? Good luck putting that genie back in the bottle.

What it means: 

Open-source AI is a double-edged sword, and China just proved it. This is gonna make companies think twice about releasing open-source models. Watch how fast new restrictions drop on AI releases.

Is Microsoft really gonna take a page out of Apple’s AI playbook? 

According to TechRadar reports, Microsoft's cooking up something interesting - potentially rebranding their AI features as "Windows Intelligence."

Internal files show they're taking notes from Apple's playbook, trying to make AI feel as natural as clicking the start button.

This ain't just a name change. Microsoft's been pushing AI into everything from Notepad to their entire Microsoft 365 suite.

But, are they really gonna do the whole “Windows Intelligence’ thang? 

What it means: 

Let’s cut through the fluff – we think this would be a bad move. 

Apple’s “Apple Intelligence” is kinda cringe to begin with, and we haven’t hesitated to call them out for thinking they can rebrand AI. lolz. 

Microsoft has a strong asset in its Copilot brand, and we think they should stick to it. 

Waiting for GPT-5 news? 

Wait no more.

(Ok, actually wait a lot longer. lolz.)

Everyone's been waiting like a year and a half for news on GPT-5, and OpenAI CEO Sam Altman just conducted an 'Ask Me Anything' session on Reddit that's got the AI world buzzing.

The tea? No GPT-5 dropping anytime soon, but he teased some "very good releases" coming this year.

Compute challenges got them working with Broadcom on custom chips, but that help ain't coming till 2026.

Also, Sam talked about how the o1 models were a huge priority for OpenAI. 

What it means: 

Sorry not sorry to everyone and their mom’s pilates teacher who kept hitting retweet on all of the GPT-5 rumors from like the spring. 

We’ve said for more than a year that: 1) OpenAI would not release GPT-5 before the U.S. election and 2) OpenAI would likely not release GPT-5 until either Anthropic, Claude or Meta releases a model that blows GPT-4o away. 

And so far…. OpenAI is still dominating the competition with a polished version of a model that’s more than a year old.