Meta uses your data to train AI, and it doesn't want you to opt out

Facebook and Instagram user data can be used to train Meta’s AI models by default. While every user has the option to opt out, Meta makes this process unnecessarily difficult. Moreover, a user's request to opt out may not be granted.

Every large language model that underpins chatbots – be it ChatGPT, Gemini, or Claude –, must be given vast amounts of data for training. In fact, the lack of training data is considered to be one of the issues that could hinder the future growth of generative AI.

Contrary to its main competitors, Meta has one big advantage: the vast amounts of data created by its users on social media.

A few weeks ago, Meta's chief product officer Chris Cox told Bloomberg that the company uses its users' data to train a text-to-image generator called Emu, which can generate an image from a prompt.

In its privacy policy, the company states that the data it uses to train AI models "could be things such as posts or photos and their captions" – basically, everything you post on the social network, except the content you share via private messages.

Over the last couple of days, Meta has notified its users that it’s planning new AI features. By default, all the users' data can be used to train Meta's AI models.

While every user has the right to opt out, this process appears to be unnecessarily complicated. Instead of allowing users to do it with a single click, Meta requires users to provide an explanation.

Moreover, it has the right not to grant your request.

"We will now rely on the legal basis called legitimate interests for using your information to develop and improve AI at Meta. This means that you have the right to object to how your information is used for these purposes. If your objection is honored, it will be applied going forward," Meta's Policy update notice reads.

The company also adds a few extra steps. A user named Tantacrul on X, formerly Twitter, says that the process is intentionally designed to be highly awkward to minimize the number of users who will object to it.

"I've worked in growth teams who conduct experiments to minimize friction for over a decade, and I know how to streamline an experience. Rule: every additional step you add dramatically decreases the percentage of people who'll make it through to the end," he said in a post.

Here are the necessary steps for those who want to opt out of having their data trained by Meta's AI models.

How to opt-out

1. In your notifications section, look for a notification where it says "We're planning new AI features for you. Learn how we use your information." It should have been sent to you in the last few days.

2. After pressing on it, you’ll see a Policy update notice. Look for the hyperlinked words "right to object" and press on the hyperlink.

3. You will see a new form with all your information that the company is planning to use for AI training. There, you have to choose your country, type in your email address, and fill in the form, providing Meta with an explanation why you don't want your data to be used for AI training.

For those who live in the EU, there is a clear reason to opt out – GDPR. I, living in an EU country, posted this text I found in an online forum.

The use of my data infringes on my rights as a (French/German/Dutch/etc.) resident and EU citizen. I do not consent to my data being used for AI training, development, improvement, or a more personalized experience. I do not give consent for my data to be used for any other form of training, development, improvement, or a more personalized experience in any Meta product or technology. I have the right to keep my data private as per the European GDPR law.

4. After you type in your request along with an email, the company will send you a six-digit One-Time Password to your mailbox. You will have to type it in the form. The company states that it will not receive submission until the code is successfully entered.

5. After that, you should get a submission confirmation. I got an email saying that my request was granted after around 10 minutes of submitting it.