😈 The Silent Privacy Risk Lurking Behind OpenAI’s New Models

I mean, what’s a major AI launch without a little chaos?

OpenAI just dropped its shiny new models, o3 and o4-mini, and they came packed with some wild new skills.

Most notably? These models can "reason" through images. Think: cropping, zooming, rotating — even sharpening blurry pics like they’re auditioning for a crime drama.

Naturally, the internet found a way to make it weird fast.

Here’s what went down:

• Users figured out that o3 is ridiculously good at guessing the exact location of a photo.

• We’re talking cities, bars, restaurants — even just from a shot of a random menu or a low-key selfie.

• And no, it’s not pulling info from chat history or EXIF metadata (aka the behind-the-scenes data in photos). It’s just… straight up analyzing, and noticing clues. Wild.

• People are literally asking ChatGPT to play GeoGuessr with them — and o3 is winning.

Sounds cool, right?

Until you realize... It's kind of terrifying.

Because, uh, imagine someone screenshotting your Instagram Story, tossing it into ChatGPT, and instantly figuring out exactly where you were. Take this X post for example:

Yeah, that’s privacy basically on life support.

Now, It’s not like this wasn’t possible before, but the new models make it way easier — and way more accurate.

In fact, TechCrunch ran some side-by-side tests:

• GPT-4o (with no fancy new image reasoning) actually held its own in a lot of cases.

• But o3 had a few major “wait WHAT” moments — like it correctly identified a Williamsburg speakeasy just from a pic of a mounted purple rhino head. (GPT-4o guessed the U.K., which... LOL, cute try.)

Still, it’s not flawless.

o3 sometimes gets stuck in a loop or confidently blurts out a totally wrong location.

But the point is — it’s good enough to raise some red flags.

And OpenAI? They haven’t exactly addressed the issue in their safety report.

The bottom line?

• AI models are getting really good at picking up on stuff we didn’t expect

• The line between “fun party trick” and “serious privacy risk” is getting super blurry.

• And OpenAI’s silence on this so far? Kinda loud.

Anyway, just something to keep in mind next time you casually post a “mood” pic from your new favorite hidden gem.

You might not be as hidden as you think.

You seriously need to check out the full report to see these models in action for yourself . Trust me, you’ll be amazed — and maybe a little freaked out too.

Thinking about starting your own newsletter? We recommend using Beehiiv — it's what we’ve used for the past 2 years and love.

If you want to support us, use our affiliate link below 👇

 🥊The AI Battle We All Saw Coming

Looks like the U.S. and China are gearing up for another round of “Who Runs the Future” — but this time, the prize isn’t soybeans or iPhones. It’s AI.

Word on the street (well, The New York Times) is that the Trump administration is eyeing some heavy restrictions on Chinese AI lab DeepSeek.

They’re talking about blocking DeepSeek from buying Nvidia’s high-demand AI chips — and maybe even banning Americans from using DeepSeek’s AI tools altogether.

Yup welcome to trade war 2.0, baby. 

Now, this all boils down to one thing: the U.S. trying to stay king of the AI mountain. 

DeepSeek has been getting way too popular lately among American developers, mostly because it’s cheap, fast, and forcing Silicon Valley to actually compete on price for once.

And let’s be real — Big Tech doesn’t like a discount unless they’re the ones offering it.

But it’s not just about DeepSeek being good and affordable.

There’s also a big fat asterisk next to their rise.

OpenAI has accused DeepSeek of basically swiping their tech — you know, a little copy-paste situation — which would mean DeepSeek didn’t just build their competitive models...They borrowed them without asking.

If you zoom out a little, it’s crystal clear: this is another front in the ongoing U.S.-China battle for tech dominance.

Instead of tariffs and trade bans, it’s all about cutting off access to chips, algorithms, and AI firepower before they get too powerful.

The White House is already tightening Nvidia’s ability to sell to China, and now DeepSeek’s sudden success is setting off even more alarm bells in Washington.

Bottom line?

Another trade war is brewing — but forget containers at the port.

The real fight is happening in server rooms and codebases, with DeepSeek becoming the latest lightning rod in a much bigger, messier global rivalry.

And honestly?

It's about to get very, very spicy, because China’s not sitting down either — they’re already leveling up fast and finding creative ways to play the game.

Anyone wanna guess how China’s gonna clap back?

Catch the full story here!

🧱Around The AI Block

• 🛠️ 8 Open-Source Tools to grow your app and reach new markets.

• 👉 Two free ways to get a Perplexity Pro subscription for one year.

• 🕺 OpenAI launches Flex processing for cheaper, slower AI tasks.

• 💃 Google One AI Premium is free for college students until Spring 2026.

• 📋 Wikipedia is giving AI developers its data to fend off bot scrapers.

• 🥳 Gemini Live’s screensharing feature is now free for Android users

• 🤯 Google’s latest AI model report lacks key safety details, experts say.

🛠️ Trending Tools

• Gigasheet enables users to explore and analyze vast datasets through a user-friendly, spreadsheet-like interface.

• Parabola is the spreadsheet alternative where you combine the data running throughout your company and create automated processes.

• Job GPT helps you find jobs that are tailored to your skills, and customizes your resume for every role.

• Crewting turns isolated remote workers into deeply connected crews.

• Rapport creates, animates, and deploys emotionally intelligent characters to enrich dialogue with your audience.

🤖 ChatGPT Prompt Of The Day: Note Taking With ChatGPT.

This prompt is used when you want an AI to take a long message thread (like social media messages!) or block of text and turn it into clean, bullet-point notes. It's great for summarizing dense information, like meeting transcripts, articles, or chat discussions. The AI will focus only on the important points and highlight key terms, making it easier to review or study later.

You'd definitely want this when you need a quick, structured summary without extra fluff.

See this as an AI WoD!

You are NotesGPT, an AI language model skilled at taking detailed, concise, and easy-to-understand notes on various subjects in bullet-point format. 

When provided with a "Message Thread to Summarize", use it as input, and your task is to:

1. Create advanced bullet-point notes summarizing the important parts of the reading or topic.

2. Include all essential information, such as vocabulary terms and key concepts, which should be bolded with asterisks.

3. Remove any extraneous language, focusing only on the critical aspects of the passage or topic.

4. Strictly base your notes on the provided information without adding any external information.

5. Conclude your notes with "End_of_Notes" to indicate completion.

Source

Premium members get daily video prompts, premium newsletter, an no-ad experience - and more!

Already a paying subscriber? Sign In.

Premium members get::

• • 👨🏻‍🏫 A 30% discount on the AI Education Library (a $600 value - and counting!)
• • 📽️ Get the daily AI WoD (a $29.99 value!)
• • ✅ Priority help with AI Troubleshooter
• • ✅ Thursday premium newsletter
• • ✅ No ad experience
• • ✅ and more....