Meta will use your old Facebook posts to train AI without your consent

Updated 22 April 2025

Meta resumes AI training on EU data – users must decide

Meta has announced that the company will now begin using publicly available data from EU users to train its AI models. This includes posts, comments and interactions with Meta AI on platforms such as Facebook and Instagram. Private messages and data from users under 18 are excluded. The decision comes after the green light from the European Data Protection Board (EDPB) and marks the resumption of a plan that was put on hold in 2024 following objections from the Irish Data Protection Commission.

All users are being informed about the changes via in‑app notifications and emails, and it is possible to object through a new form. Meta emphasises that the objection process should be easy to find and use, and that all opt‑outs will be honoured. This gives users direct control over whether their content may be used in AI training.

The rationale behind the initiative is a desire to develop AI models that understand and reflect European linguistic and cultural particularities. According to Meta, the AI should be able to interpret local humour, sarcasm and dialects, so that European users get more relevant and human‑like AI interactions. 

Meta pauses AI launch in Europe after pressure from Irish authorities

On 14 June 2024, Meta paused its plan to train its artificial intelligence on users’ personal data from Facebook and Instagram.

Read more:

Consumer Council breathes a sigh of relief after Meta postpones AI training on your old photos and posts

Meta pauses AI models launch in Europe due to Irish request

From 26 June 2024, Meta will use photos and posts from Facebook and Instagram to train its artificial intelligence. This applies to content that is publicly available, but it is unclear whether posts visible only to friends will also be used. Meta claims that it does not need users’ consent, as its interest in using the data outweighs this, which could potentially be at odds with the EU’s data protection rules.

Meta risks large fines for breaching the EU’s General Data Protection Regulation (GDPR), which can be up to 4 per cent of the company’s total global turnover.

Criticism of lack of consent and transparency at Meta

The organisation None Of Your Business (NOYB) has filed complaints in 11 EU countries because it believes Meta’s practices contravene GDPR rules. It points out that there is no clear description of the purpose of the AI training, and that users have to actively opt out rather than give consent first.

At the same time, Forbrugerrådet Tænk (the Danish Consumer Council) is investigating whether Meta’s practices are legal at all, as it is unnecessarily cumbersome for users to refuse participation.

If you do not object, your posts and photos on Facebook will, from 26 June, be fed to Meta’s AI assistant. Below, we have put together a short guide on how to prevent your data being used for the training.

Access the notice or the form

If you have received a notice on Facebook or Instagram about AI training, you can click it to access the "right to object" form.
Alternatively, you can access it directly by clicking here (remember to be logged in to Facebook or Instagram).

Complete the form:

• State your country of residence, your email address and the reason for your objection. You can use a simple statement such as "I do not want my data to be used for AI training" or provide more detailed responses that highlight your privacy concerns.

Submit the form:

• After completing the form, you will receive a one-time password (OTP) by email. Enter this OTP to confirm your submission.

Wait for confirmation:

• Meta will review your request and send you a confirmation email, usually within 24 hours, indicating whether your objection has been approved.

After submitting the objection, we received the following email from Meta:

Meta is not alone in this

Meta is not the only tech company training new models on users’ data. Below, we have compiled an overview showing that more or less everyone is already well underway:

TikTok

• TikTok uses machine learning to analyse videos and categorise the content based on, among other things, computer vision and natural language processing.
• User data such as likes, views and comments are used to train TikTok’s recommendation algorithm to personalise the "For You" feed.

What You Need to Know About the TikTok Algorithm to Go Viral in 2024

In this ultimate guide, we explain the inner workings of the TikTok algorithm, the factors that influence it, and share expert tips to boost your chances of going viral. Let’s get into it. How the TikTok Algorithm Works Before we go into the specifics of the TikTok algorithm, we need a basic understanding of how it works. Unlike algorithms on other social media platforms, TikTok’s algorithm prioritizes content based on user preferences and engagement rather than followers or popularity. Even if you’re new or have few followers, you can still go viral. This can be achieved by understanding and using

AgorapulseErica Pollock

TikTok turns to generative AI to boost its ads business | TechCrunch

TikTok is the latest company to incorporate generative AI into its ads business, as the company announced that it’s launching an AI suite.

TechCrunchAisha Malik

Snapchat

• Snapchat uses AI for object recognition, image processing, facial recognition and geolocation.
• Your conversations with Snapchat’s AI chatbot, "My AI", are used to train the chatbot’s model and to target personalised adverts.

Do you use Snapchat’s AI chatbot? Here’s the data it’s pulling from you

As Snapchat’s ChatGPT-powered AI surpasses 150 million users, here’s what data it’s gathered so far.

ZDNETJada Jones

Snapchat Aims to Spend $1.5 Billion Per Year on AI

The CEO of Snap, developer of Snapchat, says the messaging platform is increasing its spending on AI and machine learning.

PYMNTS.comPYMNTS

YouTube

• YouTube uses AI for content moderation to quickly remove offensive videos.
• User data such as watched videos, searches and likes are used to train YouTube's recommendation algorithm.

YouTube uses artificial intelligence to remove 11 million videos in a quarter

Related to violence, child safety

TechRadarRaj Narayan

The Multifaceted Role of AI on YouTube: Enhancing Content Recommendation, Moderation, Accessibility, and Creation

YouTube has become an integral part of our lives, offering a vast array of video content for entertainment, education, and more. Behind the scenes, artificial intelligence (AI) plays a crucial role in shaping the YouTube experience.

Raymond Kubode

Reddit

• Reddit has struck a multi-million-dollar deal to let tech companies use user posts and comments to train AI models.

OpenAI agrees to deal with Reddit to scrape its content for AI training - SiliconANGLE

OpenAI agrees to deal with Reddit to scrape its content for AI training - SiliconANGLE

SiliconANGLEMike Wheatley

Twitter (X)

• Elon Musk's X (formerly Twitter) has updated its terms so it can use public tweets to train its AI models, including Musk's new AI company, xAI.

Elon Musk’s xAI could train AI models on your Twitter data

With “X” aka Twitter, Elon Musk has blown a billion dollar hole in his finances. Can he plug it with the AI currency “data”?

THE DECODERMatthias Bastian

Sources

Meta will train AI on your old Facebook posts without asking for consent – the EU responds with 11 complaints

Today the EU is sending 11 personal data complaints against Facebook's parent company, Meta, which intends to train its AI technologies on years of personal Facebook posts.

ComputerworldThomas Møgelbjerg Business journalist

Facebook and Instagram want to use your old party photos to build artificial intelligence

Facebook's new plans will mean a loss of control for users, says an advocacy organisation.

TV 2 DANMARKLaura Kongsmark Schuldt

Soon Meta will use your data to train its AI: 'They are making it hard for users to object'

According to Forbrugerådet Tænk (the Danish Consumer Council), it is problematic that Meta reserves the right to reject consumers' objections.

DRAnna Danielsen Gille

Meta wants to use the data of four billion people for AI development without asking for consent | Version2

Meta's new ambitions conflict with the GDPR, stresses an Austrian privacy organisation that has complained to the data protection authorities of 11 EU countries. The company is not asking for consent, and users have less than a month to opt out.

Version2Louise Olifent

Meta Uses Your Instagram and Facebook Photos to Train Its AI—Here’s How to Stop It

Stop Meta AI from using your data.

How-To GeekAman Kumar

noyb urges 11 DPAs to immediately stop Meta’s abuse of personal data for AI

noyb urges DPAs in 11 countries to immediately stop Meta’s use of personal data for undefined “AI technology”

noyb.eu

Is Meta allowed to use your data to train its AI?

If you don't want your Facebook posts and photos to be used to train Meta's AI models, you can object. Get help here.

Home

Meta will train AI on your old Facebook posts without asking for permission – the EU responds with 11 complaints

Today, the EU is sending 11 data protection complaints against Facebook's parent company, Meta, which wants to train its AI technologies on years of personal Facebook posts.

ComputerworldThomas Møgelbjerg, business journalist

Meta faces complaints in 11 countries over data use for AI

Additional complaints will follow, according to Austrian privacy group NOYB.

EuronewsCynthia Kroet