Social Media Platforms Train Their AI Models Based on Your Content, but You Can Stop Them (Sometimes).
Amazon-owned Twitch recently made headlines by announcing that it would allow users to opt out of having their stream data, broadcast recordings, and chats used to train artificial intelligence. This, naturally, sparked outrage among many . However, it turns out that Twitch is perhaps one of the more moderate social networks in this regard. Most social media apps train some form of AI using your data, and few make it so easy to opt out with a single button.
On virtually every social media platform today, it’s extremely difficult to accurately determine how your data is being used. Training generative AI models like Google’s Gemini is often lumped in with more mundane (but still machine learning-based) functions, like YouTube’s algorithm. Reviewing a privacy policy to even determine whether a social media app is facilitating the creation of such controversial generative AI is a task that would daunt most lawyers.
Even if you know how your data is being used, many websites simply don’t provide an opt-out option. And of all the ones I reviewed, not a single one that trains AI offers a consent model. Simply put, if you visited a website today, tech companies interpreted it as your consent to use your data for training. However, if there is a way to opt out of even some AI training on social media sites, you’ll find instructions in our list below, sorted alphabetically.
Bluesky
What kind of training do they do? Bluesky doesn’t officially train any AI models , although it does use AI in its development and creates AI features . However, since Bluesky uses the decentralized AT protocol , your messages (and a lot of other data, such as who you block ) are publicly available. Therefore, virtually anyone can parse your messages and use them to train AI.
What can I do to stop this? You don’t need to do anything; Bluesky itself doesn’t use your messages to train its AI. However, if you’re concerned about third parties collecting your data, you might want to refrain from posting on this service.
What kind of training do they do? It’s easier to ask what data Facebook doesn’t use to train its AI. After spending the equivalent of a small country’s GDP on a metaverse that never materialized , Facebook’s parent company, Meta, is refocusing on AI and rushing to catch up with companies like OpenAI, Anthropic, and Google. The company’s policy on training AI using your data is extremely broad , encompassing not only your Facebook posts, photos, and interactions, but also data collected by third-party brokers and “information available on the internet.” It’s such a sweeping phrase that it’s hard to imagine what data Facebook is technically capable of collecting that it wouldn’t. Facebook says it doesn’t limit AI training to private messages with friends or family, “unless you or someone in the chat chooses to share those messages with our AI.”
What can you do to stop this? Unfortunately, with the exception of EU countries , Facebook makes it impossible to opt out of AI training based on your data. The company makes a narrow exception (where required by law) for the specific case where you discover personal information about yourself in a response from one of its AI tools. In this case, you can file a complaint , which Facebook will review and decide whether to take action.
However, it’s important to remember that Meta views interactions with Facebook’s AI tools as consent to training future AI models based on your interactions with them. Therefore, even attempting to determine whether your personal data was collected at all could lead to further vulnerability. Overall, it seems the only winning move with Facebook is not to play . Although it’s worth noting that even deleting your Facebook account won’t necessarily stop the AI from training on your data—it will simply prevent them from obtaining any data directly from your Facebook account.
What training do they undergo? Instagram is owned by Meta, Facebook’s parent company, so many of its rules are the same as those we discussed for Facebook in the section above. However, Instagram has its own unique issues, such as its Muse feature, which briefly allowed users to create AI-generated images of other users without their consent before immediately removing the feature after realizing it was a terrible, terrible idea.
What can you do to stop this? As with Facebook, you can object if you discover your personal information is included in AI responses, and if you’re in the EU , you can opt out of the training entirely. Alternatively, your only real option is to either delete your Instagram account or at least make it private .
What kind of training do they provide? LinkedIn is owned by Microsoft, and Microsoft is known to be entirely focused on AI ( using it “solely for entertainment purposes “ ), so you should be wary of how the site uses your data to train AI. According to LinkedIn’s official policy , user posts, comments, profile data, resumes, and group activity, among other types of data, can be used to train AI models. Unofficially, it’s worth noting that LinkedIn was sued last year over allegations of training AI based on private messages—a charge LinkedIn denies.
What can be done to stop this? While LinkedIn uses the same “better to ask for forgiveness than permission” model as most companies, the site at least allows you to revoke your permission. On the site, go to “Settings and Privacy” > “Data Privacy” > “How LinkedIn Uses Your Data” > “Data to Improve Generative AI.” Here you’ll find a toggle that’s enabled by default and labeled “Use my data to train AI models to create content.” Turn it off. Note: This only applies to data collected within LinkedIn, not the broader Microsoft ecosystem.
What kind of training do they do? Reddit is complicated: the company doesn’t train its own AI models, but it has agreements with OpenAI and Google to train their models on its data. Reddit has reportedly at least considered terminating these agreements (at least in part because, like everyone else, it’s seeing traffic decline as AI displaces search engines). But AI models were also trained on Reddit data long before these agreements became official, so it’s difficult to say whether ending official partnerships will hinder AI companies’ ability to collect as much public data as possible.
What can be done to stop this? Short of deleting your Reddit account and completely stopping posting on the site, there’s nothing you can do. Reddit currently doesn’t have a tool for opting out of AI training. And even if it did, AI companies have long been harvesting publicly available posts.
Snapchat
What kind of training do they provide? Like most social media apps, Snapchat uses your publicly shared images, videos, and audio to train its generative AI models. The company says it uses them to develop features like AI Snaps and AI Lenses. Snapchat also had a short-lived agreement to integrate Perplexity’s AI tools into its search engine, though this agreement was ” amicably terminated ” last year before widespread adoption, so it’s unclear whether user data was ever shared with Perplexity or used for training.
What can I do to stop this? Unlike most social media apps, opting out of generative AI in Snapchat is fairly simple (albeit a bit hidden). Within the app, open “Settings” and under “Privacy Controls,” tap “Generative AI Settings.” Here, toggle off the “Allow Use of Public Content” toggle. That’s it. As usual, third parties can collect public data, though Snapchat’s design makes this a bit difficult. If you want to be completely safe, avoid leaving any permanently accessible public content on your profile.
Threads
What kind of training do they do? Again, Threads is owned by Facebook’s parent company, Meta, so it has the same data training rules as Facebook. In other words, unless absolutely required by law, publicly available data on Threads will be used for AI training regardless of your consent, and you cannot opt out.
What can you do to stop this? If you’re in the EU, you can file a complaint using the same procedure as for your Facebook account. US users will need to file a complaint confirming that your personal information appeared in AI responses to have some of your data removed. Beyond that, you have only one option: either make your Threads account private or delete it entirely .
TikTok
What kind of training do they do? TikTok’s AI training policy is an even bigger nightmare than Facebook’s, due to the company having to spin off its US division earlier this year . Generally, TikTok’s parent company, ByteDance, trains AI models on user data. While this data can be used for other purposes in the US, such as training recommendation systems, it cannot be used for more general training of ByteDance’s AI models.
What can you do to stop this? As with Facebook, if you want to opt out of AI training on your data, you need to file a specific privacy objection in accordance with local laws. In other words, you can’t simply say, “Please don’t train AI on my data.” You can visit this page to access the appropriate form for your region.
Normally, in this case, we’d say closing your account is a good backup measure, but there have been reports of TikTok even training AI on private videos and unsaved drafts. The most reliable way to ensure TikTok doesn’t train AI on your data is to delete your account.
Twitch
What kind of training, exactly, do they provide? Twitch’s policy allows for the use of your streams, broadcast replays, clips, chat messages, and even images and text on your channel to train generative AI models. It’s unclear what specific features Twitch will use this data for. The company mentions features like subtitles on its FAQ page , though it’s clear that automatic subtitle generation existed long before the advent of generative AI models as we know them today.
What can I do to stop this? If you have a Twitch channel, you can go to your Twitch safety settings , scroll down to the “Generative AI Training” section , and turn off this toggle. Yes, it’s on by default, and yes, Twitch knows it’s unpopular . It’s important to note that this setting will only apply to your channel. If you appear on someone else’s stream or post in their chat, their opt-out preferences will determine any data you generate.
X
What kind of training do they provide? Not only does X use your public posts and data to train its Grok AI, but since the company has teamed up with xAI and SpaceX, your data may be used to train AI models across the company. I can only judge this based on the site’s public policies, though it’s worth noting that SpaceXAI CEO Elon Musk is known for his disregard for legal and even security procedures . So, take this into account when assessing the risks when deciding what to do with your data.
What can you do to stop this? Luckily, X has a way to opt out of training AI on your data. Go to Settings > Privacy & Security and scroll down to the Data Sharing & Personalization section. Here, select “Grok and Third-Party Partners.” On this page, you’ll see several checkboxes you need to uncheck, including the detailed option “Allow Grok and xAI to use your public data, interactions, input, and results for training and fine-tuning.”
YouTube
What kind of training do they do? YouTube is a privacy nightmare when it comes to AI training, as your YouTube account is linked to your broader Google account, and Google is one of the biggest players in AI . It would take a much longer and more detailed guide to detail all the ways Google trains AI on your data (even outside of your Google account) and how to stop it. But even within YouTube, Google has been found to be training video generation models on videos uploaded to the platform , without the knowledge or consent of YouTube users.
What can you do to stop this? If you regularly use YouTube, you should consider disabling Google’s artificial intelligence training on your data (as much as possible) at the Google account level . However, if you upload videos to YouTube, the situation is different. Google currently doesn’t offer any options to completely prevent its AI models from training on your videos.
However, paradoxically, Google provides the option to opt out of training third-party models using your data. To do this, open YouTube Creator Studio, go to “Settings” > “Channel” > “Advanced Settings,” and then scroll down to the “Third-Party Training” section . Here, you can uncheck “Allow third-parties to train AI models using my channel’s content.”