• The Automated
  • Posts
  • 🚨Google’s AI-Training Secrets Exposed!

🚨Google’s AI-Training Secrets Exposed!

Plus, a quick guide to picking the perfect OpenAI model for your needs

In partnership with

Hello and welcome to the Automated, your AI tour guide.

Here’s a quick reminder 😊

Starting May 31, 2025, our premium plan will go from $9.99 to $19.99/month.

But here’s the good news: if you upgrade now, you can lock in the $9.99 rate with any monthly or annual plan—plus you get a free 7 trial!

So don’t wait, go ahead and secure your rate today!

More info here.

Here's what we have for you today

 šŸ¤Æ Google’s Sneaky AI Moves 

So, Google has this thing where publishers can opt out of having their content used to train AI models, right?

Well, recently, we found out that while that seems to apply to Google's DeepMind team, it doesn’t stop other teams—like Search—from using the same content to train their own AI tools.

And guess what?

This came out in court on Friday when a Google VP, Eli Collins spilled the beans.

And, honestly, publishers aren’t happy about it.

Here’s what’s really been going on behind the scenes:

Turns out, even though publishers can opt out of having their content used to train Google’s AI, that ā€œopt-outā€ doesn’t really cover everything.

Specifically, it only applies to Google DeepMind, the research lab behind the AI tech.

But when it comes to Google’s Search team, they’re free to use that same content to train their AI for features like AI Overviews—you know, those little boxes that show answers right at the top of search results.

But here’s the problem: 

Those AI-generated answers you see at the top of your Google searches? They’re built using content from all over the web—including the stuff from those ā€œopted-outā€ sites.

And because those summaries are front and center, people are getting their answers straight from Google without ever clicking on the actual websites.

Which means publishers who thought they were off the hook are now losing clicks.

And fewer clicks = less revenue. So, not only is Google using their content anyway, it’s also kinda killing their traffic and revenue in the process.

Honestly, it’s a bit like lending someone your favorite book, only to find out they made a ā€œsummaryā€ of it, slapped their name on the cover, and now no one’s reading your original.

And get this: Publishers can only decline having their data used in search AI if they opt out of being indexed for search, Google clarified.

In short, Google says they respect opt-outs, but when it comes to training their AI for search, they’re basically like, ā€œThanks for the data, but we’re still gonna use it.ā€

So where does this leave the publishers?

Well, that’s a whole new headache, and we can totally see a lawsuit coming, one after another.

Really, both Google and the publishers are about to go at it in a full-blown battle.

Here are the full details of what went down in court on Friday.

šŸ”§ AI Troubleshoot: When ChatGPT Gets the Facts Wrong

 Hey Automaters, we’ve got another fix coming your way!

One of our readers reached out after getting outdated info from ChatGPT — and honestly, we’ve been there too. You ask a simple question, and it spits out something that sounds right… until you realize it’s a few months (or years!) behind.

So what’s going on?

Well, here’s the deal: unless ChatGPT has browsing turned on, its knowledge is frozen as of December 2023. So if you’re asking about something recent — new tools, live news, or last week’s tech launch — there’s a chance it’s giving you outdated info.

Here’s how to get more accurate answers:

  1. Check the Knowledge Cutoff: GPT-4’s default model doesn’t automatically know what happened after Dec 2023. If you’re not sure how up-to-date it is, just ask: ā€œIs this based on your latest knowledge?ā€

  2. Enable Browsing (Pro/plus users only): If you're on the Pro or plus plan, you can turn on browsing in Settings > Beta features > Browse with Bing. This lets ChatGPT pull in fresh info from the web.

  3. Use a Custom GPT with Browsing: Some Custom GPTs come with browsing built in. Look for ones that say they support live data or recent updates — it’s a great workaround for time-sensitive tasks.

  4. Double-Check the Facts: If you’re dealing with anything time-sensitive or important (like financial data, legal info, or breaking news), always cross-check with a trusted source like Google, DeepL, or official sites.

That’s it for this round!

If you're out there dealing with an issue that's driving you nuts, reach out to us here, and we’ll go out of our way to help you.

P.S. Premium subscribers get priority support!

Now of to the next big news…..

 šŸ¦¾ How to Pick the Perfect OpenAI Model for Your Needs

If you've been scratching your head wondering which OpenAI model to choose, good news—they’ve just dropped a guide to make things crystal clear.

Yep, they quietly posted an article titled "ChatGPT Enterprise - Models & Limits" on May 3rd, where they broke down the five models on offer and gave us the lowdown on which one to use and when.

And while it's mainly geared toward businesses, regular users like us can totally benefit from it too.

So, let’s talk about the 5 models you need to know about:

  • GPT-4o: This is the ā€œomni modelā€ā€”the one you want for real-time info. Think of it as your all-in-one Swiss Army knife—perfect for brainstorming, summarizing content, drafting emails, and much more. Plus, it’s multimodal, which means you can dive into custom GPTs, image generation, audio, and even data analysis. Basically, it does everything—and then some.

  • GPT-4.5: If you’re looking for a bit more flair, creativity, and emotional intelligence, GPT-4.5 is your go-to. It’s got that extra creative touch, especially for brainstorming or unlocking those ā€œout-of-the-boxā€ ideas. If you're not just looking to get things done but want to feel something along the way, this is your model.

And about the mini models?

Things get a little more specialized here. OpenAI’s mini models are great at handling specific tasks, each with its own sweet spot.

Here’s what they’re about:

  • o4-mini: This one's the cost efficient fast runner. If you’re all about quick technical tasks like programming, STEM-related queries, or just something that needs to be done fast and snappy—o4-mini is your buddy.

  • o4-mini-high: This is the overachiever. If you’re deep into complex coding, advanced math, or scientific explanations that require high accuracy, o4-mini-high is your best friend. In fact, it’s the ā€œhigh-endā€ version of the o4-mini family!

  • o3: Here’s where the heavy lifting happens. If you’re tackling complex, multi-step tasks like strategic planning, massive coding projects, advanced math, deep science, or analyzing big data sets, o3 is your all-star model for the job. It's perfect for when you need careful planning and detailed, heavy-duty brain work.

Oh, and here’s the catch:

You’ll need the $20 ChatGPT Plus plan to unlock all these models. But once you’re in, the options are there for the taking.

  • For quick tasks? Go mini.

  • For creativity and emotional depth? GPT-4.5 has your back.

  • For complex, multi-step problems? o3 is the heavy hitter.

Now it’s all clearer, right?

Go ahead, pick the right model that fits the job!

P.S. There’s lots more where that came from — click here to get the full report.

Use AI as Your Personal Assistant’

Ready to save precious time and let AI do the heavy lifting?

Save time and simplify your unique workflow with HubSpot’s highly anticipated AI Playbook—your guide to smarter processes and effortless productivity.

🧱Around The AI Block

  • šŸ„‡ The best custom GPTs to make ChatGPT even more useful

  • 🤯 One of Google’s recent Gemini AI models scores worse on safety

  • šŸ¤ Apple teams up with Anthropic to build an AI-based coding assistant.

  • šŸ§’ Google will put Gemini AI in the hands of kids under 13. 

  • 🄳 Alexa Plus now has more than 100,000 users.

  • šŸ‘Øā€āš–ļø Activist Robby Starbuck sues Meta over AI answers about him.

  • šŸ’ƒ Google’s NotebookLM Android and iOS apps are available for preorder. 

šŸ¤– ChatGPT Prompt Of The Day: When You Want to Change that One Word in Your Writing?

To execute this prompt effectively you need to:

  1. Always Include Context: Don’t isolate the word—share the full sentence or paragraph so suggestions are accurate.

  2. Clarify Your Intent: Mention the tone, emotion, or connotation you're aiming for. (e.g., "I want a word that sounds more formal/sarcastic/poetic/etc.")

  3. Specify the Target Word: Clearly point out which word you want to replace.

  4. Mention the Audience or Style: Let the AI know who you're writing for—this greatly affects word choice.

Below, you can see the full prompt and how it works live as an AI WoD!

That's all we've got for you today.

Did you like today's content? We'd love to hear from you! Please share your thoughts on our content belowšŸ‘‡

What'd you think of today's email?

Login or Subscribe to participate in polls.

Your feedback means a lot to us and helps improve the quality of our newsletter.

šŸš€ Want your daily AI workout?

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

Already a paying subscriber? Sign In.

Premium members get::

  • • šŸ‘ØšŸ»ā€šŸ« A 30% discount on the AI Education Library (a $600 value - and counting!)
  • • šŸ“½ļø Get the daily AI WoD (a $29.99 value!)
  • • āœ… Priority help with AI Troubleshooter
  • • āœ… Thursday premium newsletter
  • • āœ… No ad experience
  • • āœ… and more....

Reply

or to participate.