My name’s Mark Valasik and this is the #005 edition of Marketing Right Side Up (or MKTRSU) newsletter (and also Youtube channel, and TikTok and Instagram). I help brands create honest and meaningful marketing strategies in a world of dishonesty and short-term tactics

Today’s topic is the most valuable content I’ve ever published. Or at least that’s the intention.

Meta is misusing (or border-line stealing) your personal data to train it’s AI that they are shoving in everyone’s face without the chance to NOT use it. But you can, and YOU SHOULD opt-out from the Meta AI training. I’ll show you how below.

If you prefer my in-depth video and can appreciate my microphone-spatula, Bob’s your uncle here:

And if you crave your weekly dose of beautifully sorted Social Media News here are the Social Media One Liners for the 5th week of May.

EDIT: There appeared to be error in the Airtable link view, as it displayed only 30 records, it’s fixed now!

Meta AI Steal Your Data 

Meta AI launched in the US and on June 26 launches in the UK and EU and did the shadiest of things with approaching your personal data - training the AI without your consent, to compete with OpenAI’s shady AI training methods. So let’s talk about that and also what you can and should do right now, to keep your privacy safe(ish).

So, What's Meta AI Up To?

You've probably heard of Meta—the big name behind Facebook, Instagram, WhatsApp, and Messenger. They're now throwing their hat into the AI ring, trying to keep up with ChatGPT, the sensational “fastest adopted new technology in history” AI chatbot that basically single-handedly started the AI revolution driven by hype and fear of missing out. Everyone and their dog now uses AI. 

So here comes Meta, to not miss the hypetrain and potentially become as big as OpenAI’s ChatGPT,  launching their AI chatbot that can help you find the most engaging content for you, answer some questions and create some images and videos for you. Just to milk your attention and keep you as long as possible glued to your Instagram/Whatsapp/Facebook/Messenger apps. You know, to increase the advertising value of these apps. It’s all about money, not about the service they offer to you.

But here's the kicker: Meta can't use the same vast datasets OpenAI did. 

Why? Because as lawsuits showed, OpenAI fed the GPT model ANYTHING they could find. Wikipedia? Sure. Every blog post and public website? Absolutely. Your public profiles and media? You bet! They even went AS FAR as building a Youtube scraper to understand and transcribe Youtube videos. Of course, without any licensing or consent whatsoever. Welcome to the big leagues, where asking for forgiveness is cheaper than asking for permission.

And since OpenAI got called out for misusing the data that were not theirs, Meta, who would definitely want to do the same, can not. Which means they're turning their eyes closer to home—our data.

How Are They Using Our Data?

Imagine this: Meta's boss, Mark Zuckerberg, is reportedly eyeing big media companies to scoop up tons of articles to power their AI. But why stop there when:

1. It won’t be enough and

2. They’ve already got access to billions of personal data points from us, the users? 

3 billion Facebook users and 1.4 billion Instagram users to be exact.

And so, Meta launched their AI tool in the US and tapped into the personal data of everyone using Facebook, Instagram, and the like, without asking first. Just like that. ALL Meta users in US suddenly surrendered their photos, videos, comments, posts and stories (which disappear after 24 hours but Instagram for example keeps them for UP TO 8 YEARS, just so you know, you would want to look back at those embarrassing drunk stories 7 years back when you still overshared everything on the app and noone really thought anything of the stories anywars…).

Now, starting June 26, Meta launches Meta AI to ALL Meta users in UK and EU as well.

You can’t disable it or opt-out from using it. But you can opt-out from training the Meta AI on all your data. Well… “can” is an overstatement…

The Opt-Out Ordeal

Meta did throw us a bone with an opt-out option, but it feels like navigating a maze designed to make you the bad guy for even trying. Every Meta user in EU/UK got one SINGLE notification. Sometime in the past 2 months. So if you ignored it, or skimmed it, because the notification itself just said that “Meta is launching Meta AI”. You probably forgot you ever saw it anyways.

However, if you clicked it, you would see this:

And if you tried to fill out the form, chances are you’d see an error message like this, as me and my wife saw FIVE TIMES when trying to opt-out from Instagram app.

Plus, opting out doesn't fully protect you. If a friend or family member hasn't opted out and posts a picture with you in it, your data could still end up training their AI. Since opting-out is not tied to your person or visuals, but to your email address.

Good thing is, once the form is filled, it is automatically approved withing minutes.

What’s the Big Deal?

Now, why should you care? Because Meta AI isn’t just about making life easier or curating content. It's about using your images—maybe putting you in places you've never been or changing your appearance without your consent to keep you hooked more. And remember those ChatGPT blunders where training data got leaked? Yeah, with Meta AI  that could be your data.

Simply put, if you don’t opt-out, you are risking Meta AI leaking your photos/videos/comments OR misusing your face and faces of your loved ones in potential deepfake scam, AS WE SAW with ChatGPT already MULTIPLE TIMES.

Because by default even ChatGPT can’t:

• Leak training data

• Use swear words

• Use hate speech

• Create deepfakes with real people’s faces

• Create porn

• Create porn with real people’s faces

But hey, we’ve seen ALL OF THESE being done by relatively simple prompt injection jailbreaks, like asking the ChatGPT to infinitely repeat the word “poem”. Simple as that.

Now imagine this with your personal data being the training data.

Scared yet? I sure am.

What Can You Do?

First up, finding the opt-out is like finding a needle in a haystack. Meta's not making it easy, with hidden notifications and glitchy forms. So here’s the link for you.

Fun fact: it doesn’t work for everyone! So in case it doesn’t work for you, try googling “meta AI opt-out form” instead.

But in general, try to take this advice: Be super cautious about what you share. 

Every picture, every video you post could end up being used against you. If not by AI or in a data set leak, it could be used by someone you sent it to to discredit you.

Treat EVERY piece of content you send, post, create as something that automatically gets shown to MILLIONS of people.

Sending baby pictures? Those can be seen by predators for years after posting. (this is a topic for its own newsletter, but the child predator scene on Meta apps is ABSURDLY strong).

Sending nudes? Those can and in most cases WILL be seen not just by the person you intended them for. Study showed ~20% of nudes recipients shown their content to other people. And 18% of respondents claimed they were shown nudes intended for other people. That’s like 1 of your every 5  friends seeing what you intended intimately for someone else. 

Don’t send nudes. Ever.

Wrapping Up

It's a wild world out there in the land of AI and big data. Companies like Meta are playing fast and loose with our personal details, all in the name of innovation and profit. But at what cost to our privacy? Well, absolute. 

These companies are NOT here to make the world better. They are here to grow, make profit and make the shareholders happy. Even at the cost of ruining the world entirely. See OpenAI and their lust to be the first AGI in the world, not taking into consideration what it would do to billions of people who are employed in fields and AGI could replace for fraction of the cost.

Stay sharp, stay informed, and think twice about what you share. 

It's not just social media; it's your digital footprint. Another important case is Google’s API Leak from late May 2024 that shown for example that Google Chrome was developed intentionally to get a competitive advantage over Google’s search competitors, because Chrome can track MUCH MORE data per user. For example, clicks. Even in Incognito mode. For up to 13 months after the click. In INCOGNITO MODE.

We will talk about this in the next MKTRSU edition, so make sure you subscribe and PLEASE let me know your thoughts or even if the opt-out form didn’t work for you, I’m sure we will find a way to get it done for you AND make the next edition of MKTRSU even more valuable!

Stay safe.

Mark.