Every time you ask an AI assistant a question or upload a photo to social media, that content can ultimately help improve a model. This is usually outlined in the terms of service you agreed to when you created your account.
The good news is that almost all platforms let you choose, and finding each setting only takes a few minutes. We’ll tell you where each one is and what rights protect you so you can decide for yourself.
AI and data protection: why are your photos and texts used to train algorithms?
An AI model learns by example, like someone who reads many books to improve their writing. The more varied texts and images it sees, the better it understands how we speak and what the world is like, and that’s why companies seek out material to improve them.
Some of that material comes from the website itself, and that’s where automatic tracking comes into play.
Massive “scraping”: how Big Tech companies collect your digital footprint
Web scraping is the automated crawling of public websites that copies their text and images in large quantities. Your digital footprint, which includes everything you publish and share publicly, can be part of what it finds.
According to their policies, tech giants also use public posts on their platforms and some conversations with their assistants. Fortunately, almost all of them now explain how to say no, and the law supports that right.
Your rights under current regulations and AI law
The General Data Protection Regulation (GDPR) allows you to object to the processing of your data and request its deletion. This objection is crucial when dealing with companies like Meta or LinkedIn, which rely on legitimate interest to use publicly available content.
The AI law adds transparency, because providers of general-purpose models must publish a summary of the content they use to train them. With that framework clear, let’s move on to the adjustments, platform by platform.
Step-by-step guide: how to disable AI training in your usual apps
The paths may vary depending on the version of each application. The names of the settings, however, are usually the same.
1. OpenAI (ChatGPT): How to delete your model training history
Open Settings, go to Data Controls, and turn off “Improve model for everyone.” From then on, your new conversations won’t train OpenAI’s models, and saved chats will remain in your history.
For a one-off inquiry, the temporary chat is not used to improve models and is deleted after 30 days. OpenAI also offers a privacy portal with the option “Don’t train with my content.” Those who share photos and statuses on social media will find the following setting in the Meta apps.
2. Meta (Instagram and Facebook): the form to object to the use of your data
Meta uses publicly available content from adults in the EU for its AI and, according to the company, does not use private messages. To opt out, go to the Facebook or Instagram Privacy Center and look for “Right to Opt Out.”
A simple form opens where you can explain, if you wish, how the treatment affects you. Google manages this control differently, with its own settings in your account.
3. Google and Gemini: disable storage of your web and voice activity
Want to stop Gemini from saving your chats? Look for “Gemini app activity” and turn it off. With it disabled, your chats won’t be used to improve Google’s models, and you can also delete past activity.
The company retains this activity for up to 72 hours to ensure the service functions correctly. In your Google Account, under Data & Privacy, you can also pause web and app activity, as well as voice and audio. Microsoft has its own settings, spread across various services.
4. Microsoft and Copilot: Privacy settings in Windows and professional networks
In the Copilot app, open Settings, Account, Privacy, and Model Training, and turn off both text and voice training. If you opt out, those conversations will no longer be used to train Copilot models.
On Windows, optional diagnostic data is disabled under Privacy and Security. LinkedIn, which belongs to Microsoft, has its setting under Data Privacy, within “Data to improve generative AI.” One last group remains: design tools.
5. Adobe and Canva: prevent them from using your projects and designs without your consent
Adobe states that it has never trained Firefly on customer content and that it uses licensed material, such as Adobe Stock. Its terms allow it to analyze your content using machine learning, and you can disable this in your account privacy settings, under “Content Analytics.”
In Canva, your Privacy Settings determine whether your content and activity are used to improve its AI features. Review your Privacy Settings, as it only uses them when permitted by those settings.
What if your data was already used before you changed the settings?
What to do if a company has already used your data without permission?
Start by writing to the company. You can request in writing that they object to and delete your data, through their privacy email or their data protection officer, and the Spanish Data Protection Agency (AEPD) publishes request templates.
The company must respond within one month. If they do not respond or their response is unsatisfactory, you can file a complaint with the Spanish Data Protection Agency.
Keep in mind that the opposition takes effect from the moment you exercise it, so it’s best to file it as soon as possible.
Conclusion: Take control of your digital privacy against AI
Protecting your data and using AI are compatible. The settings we’ve discussed allow for the everyday use of these tools, and European regulations support your decision.
We suggest you review them from time to time, as platforms frequently update their policies. By doing so, AI can be used more safely and your data stays where you want it to.