In the digital world, trust is oxygen. When you entrust a service with your data, you’d expect it to be kept private and secure. That’s the position pCloud has held from day one. Your files are yours, and your privacy is non negotiable.
Unfortunately, not every company shares our philosophy. As AI becomes the new engine powering everything from search results to photo sorting, many tech giants have quietly rewritten the rules of data use. Instead of asking for explicit permission, they treat your personal information as part of their default AI training set. Sure, some offer opt outs tucked deep in their account settings. And perhaps we should thank them, because others don’t even bother.
So let’s pull back the curtain and examine who’s helping themselves to your data, and what you can do about it. We’ll begin with the more problematic cases and continue with platforms which have clear opt out flows.
If you are a Tik Tok user, your data is automatically available for AI to train on it. There’s no opt out toggle or way to object. Well, you can always object by not using the platform. Or you can potentially limit the future data available, by making your profile private and deleting past videos. However, what’s done is done.
Similarly, anything you post publicly on Reddit is licensed data used for AI model training. There’s no opt-out setting to prevent this.
Snapchat uses your data to train AI models through the “My AI” chatbot and its generative AI tools.
Unlike regular chats with friends, all messages with “My AI” are retained unless the user actively deletes them. Snapchat explicitly states that these retained messages “may be used to improve Snap’s products and to personalize your experience, including ads.” This means that, by default, your conversations with My AI are used to feed and improve Snap’s AI models and personalise your advertising profile.
There is no “opt-out” setting to prevent future conversations from being retained. However, users can delete their past “My AI” data. It takes about 30 days to have your data deleted from their servers. However, probably, it’s already been used for AI training.
As of Dec. 16th, 2025, Meta’s updated Privacy Policy will allow the company to harvest both your public content, and your AI chat interaction, as training data for its AI systems. According to Meta, they do not use direct messages or private chats for training. At a glance, that might seem reasonable. Public content is public after all, and private messages are off-limits.
The problem is that there’s no simple opt-out option. Users must choose between deleting their accounts, submitting a difficult-to-prove objection, or simply being mindful about their posts and AI chatbot interaction. Another caveat is that not all of these options are available to everyone. For users in the EU, Meta theoretically offers the right to object, thanks to stronger data-protection laws. But even for them, opting out isn’t straightforward. For users in the U.S. there’s no objection option. The choice is either accept Meta’s training, or delete their account.
How to submit an objection on Facebook
Step 1: Go to the Privacy Center
Open the Facebook app:
→ Tap Menu
→ Scroll down to Settings & Privacy
→ Privacy Center
Step 2: Access the Right to object form
In Privacy Center:
→ Tap Menu (top right three line)
→ Select Other policies and articles
→ Go to How Meta uses information for generative AI models and features
→ Scroll to the Right to object tab
→ Click on Learn more and submit requests here
Step 3: Submit your objection
Check the box at the bottom → Complete the form → Submit
How to submit an objection on Instagram
The process here is pretty much the same. If your Instagram and Facebook accounts are linked, you only have to send one objection.
No What’sApp objection option
Currently, there’s no direct way to submit an objection request through WhatsApp. There used to exist an option to disable the feature, but it’s been removed. Meta claims that your personal messages are off limits and only your AI chats are being used to train the model. If you are serious about your privacy, all you can do is refrain from using the chat bot.
Elon Musk has, of course, an AI of his own. Grok AI is training on your data by default. However, we’ve tested some EU accounts which did not have that set by default. In any case, wherever you may be from, it’s best to follow the steps below to ensure your settings are set to your preferences. Again, note that opting out now won’t remove past data from the model.
How to opt out
Go to Settings and Privacy
→ Privacy and safety
→ Grok & Third-party Collaborators
→ Uncheck the boxes or turn off the toggle
Since Nov. 3rd, 2025, LinkedIn has been training AI models with your public profile, resume, posts, comments, and more. The company uses opt in by default and also inverts the consent relationship. If you wish to ensure that your future data is not part of the training, here’s how to opt out.
How to opt out
Go to Settings & Privacy
→ Data Privacy
→ Data for Generative AI Improvement
→ Toggle off Use my data for training content creation AI models
Starting Oct. 10th, 2025, Google’s Gemini AI has been scanning, reading, and analysing your mails and their attachments. To stop your data from helping with Gemini AI’s training, first turn off Gmail’s smart features. Google notes that regular Gmail functions don’t contribute to training the main Gemini model.
How to opt out in Gmail
Open Settings (gear icon)
→ See all settings
→ Scroll to Smart features and personalization
→ Uncheck the box
→ Save the changes
Open Settings (gear icon)
→ See all settings
→ Scroll to Google Workspace smart features
→ Expand the section
→ Turn off toggles
→ Save the changes
How to opt out in the Gemini app
Go to Gemini Apps Activity → Tap on Turn off → Accept the terms
Even if you disable Gemini Apps Activity, Google may still use your conversations to generate anonymised data for improving its services. They know how to twist language, so your information remains available to enhance their ecosystem.
Many large language models are trained on the Stack v2 dataset. It’s a collection of public source code in 600 programming languages totalling about 67 TB. It contains code from GitHub repositories with permissive licenses.You’re able to opt out so your public repositories aren’t included or used for training.
How to opt out
Open the Menu
→ Settings → Copilot dropdown → Features → Scroll down to Privacy → Disable
Pinterest uses your data for two primary purposes that require an active opt-out. First up, it trains its image generation model Pinterest Canvas on certain publicly available data. This setting is generally considered opt-in by default. The platform also uses your activity on and off the platform to build a profile for personalised ads and recommendations. You can control which data sources are used for this personalisation.
How to opt out from both
Go to Settings → Privacy and data → Uncheck the relevant boxes → Save
Adobe’s updated terms now allow the company to use your creations to train its AI models. In other words, anything you make in apps like Photoshop or Illustrator could help power tools that might compete with your own work. Here’s how to prevent Adobe from using your artwork to train their AI models.
How to opt out
Go to Privacy Settings → Content analysis → Turn the toggle off
Medium has taken a strong stance against the use of its content for training AI models without explicit consent. The platform’s policy is “Default No” for content published on the platform. Medium’s official policy is to block AI companies from training on stories published on the platform. They have updated their Terms of Service and use technical measures (like robots.txt blocks) to prevent unauthorised scraping by AI crawlers.
***
When a company uses “opt-out” rather than “opt-in,” it inverts the consent relationship. The burden is 100% on the user to police their privacy. If you miss the window, mis-click, can’t find the form, if the form doesn’t work, or doesn’t even exist, it’s permission by default. Even if you succeed, the “removal” of your data is theoretically possible but practically uncertain. In many cases the data has already been scraped and absorbed into AI training sets. For users, especially casual ones, the chances of truly protecting their data are slim. Well, unless they make privacy a project.
Companies may argue that this approach is necessary for service improvement and innovation. However, it fundamentally conflicts with the principles of data privacy by design, user autonomy, and meaningful consent. All things held almost sacred by major global regulations. The kind of commitment pCloud tries to embody is that your data stays yours unless you choose otherwise, and not because you forgot to click “object.” We need to shift toward a default opt-out, coupled with greater transparency.
