Protecting your work from AI training: why it matters and how to take control

This guide provides general information only and is not legal advice. For advice on specific circumstances, you should consult a qualified lawyer.

Why are platforms changing their rules on using your data?

More and more platforms are quietly rewriting their Terms and Privacy Policies so that they can reuse what you upload - text, images, CVs and portfolio information, artwork, whatever you publish - to train AI systems. These changes may affect both your personal data (where GDPR applies) and potentially your copyright, although these are separate legal issues. 

Moving toward “we’ll use your data for AI unless you object,” not “we’ll ask first”

LinkedIn, Meta (Facebook/Instagram), Pinterest, WeTransfer and other platform services - widely used by artists, creators, gallerists and others who share artworks online - have all changed their Terms and Privacy Policies to enable the use of user content (which could include images of your copyright-protected artworks) and personal data for AI-related purposes. 

For many of these platforms such as LinkedIn, this will now be enabled by default in regions including the UK, unless you object to these uses. 

Meta has taken a similar position saying that it can use public content (that could  include images of your copyright-protected artworks) and can include personal data from adults’ Facebook and Instagram accounts to train its AI systems under “legitimate interest” rather than asking for explicit consent up front. Facebook, Instagram and LinkedIn do offer a “right to object” form for some users (for example, in the UK and EU), but the onus is on the individual user to find and submit that form yourself. 

At the same time, file-sharing and creative tools have also updated their policies. WeTransfer’s recent terms caused alarm among artists because the wording appeared broad enough to allow analysis of uploaded work, although the company later issued clarifications saying it does not train AI on user files. The language WeTransfer used to communicate to users had already triggered serious concerns about how creative work - which is usually copyrighted - can be repurposed without clear consent, control or licensing. 

In summary: many platforms are moving toward “we’ll use your data for AI unless you object,” not “we’ll ask first” which impacts how your personal data is used and also how your content which may be copyright protected, is used to train AI models and programmes. 

Why is AI training a concern?

Transparency

Platforms are changing how they use data and content, which in the case of artists – could include their copyrighted work and separately, personal data. Some of these changes relate to AI training, without making this clear at the point of upload. Vague terms like “improving our services” or “developing new technologies” can be broad and may include AI or machine learning uses, even if this is not explicitly stated. This could mean that your content that might be copyrighted, and/or your personal data may be used to train AI tools without giving meaningful informed consent about these specific uses.

Loss of control

Your work is your copyright or Intellectual Property (IP). Copyright protects creative works, while data protection law (like GDPR) protects personal data. When platforms say that they can analyse your posts, captions, images, text or tone of voice to “train AI,” they are claiming the right to learn from your work or personal data - often without explicit consent, credit or payment. This may undermine both your personal data rights and copyright. 

Burden on artists, creators and estate managers

Many platforms now treat your work/content or personal data as available for AI training unless you say no. That shifts the burden onto you to find and use opt-out tools. Those tools mainly address personal data under GDPR and don’t substitute for copyright licences on your creative works. Because platforms are increasingly including AI training rights through their terms of service, opting out may be the only practical way for individuals and organisations to show they do not want their content or personal data used to train these AI models.

Read more about how DACS is supporting artists and their beneficiaries as AI develops to ensure their rights and royalties are protected.

Why does this matter for artists, creators, gallerists and others who might share artworks on social media and other platforms?

For individual artists:

• Your work (which may be copyrighted) could be used to train AI models through content you post depending on the platform’s terms. An image, part of an image or creation, any writing about your artistic practice that you upload could be analysed to train AI systems. Those systems might generate outputs that resemble aspects of your works, style, tone or visual language without credit, control or payment.
• You could lose control over how your works, creative styles or voices are reproduced. For example, LinkedIn describes its AI training as helping with “content generation,” which might include drafting posts or suggestions. This means a professional tone or creative approach could contribute to tools that imitate aspects of a style without attribution or payment. For some individuals, this may raise copyright and/or personal data concerns. 

For organisations (Galleries, Auction Houses or others):

• Your organisations’ public communication such as artwork sales, curatorial texts and artist information can all become training data. For example, LinkedIn explicitly lists things like posts, and other activity as material that may be used to develop “content-generating AI.”
• Your content could be used to train tools that imitate an artists’ style, tone or creative output without an artist or artist estate being asked, and without attribution or payment. For some artists or estate managers, this may raise concerns related to copyright and/or personal data. 

Can I opt out?

Some platforms offer separate controls for personal data and for content you’ve posted.

For personal data, you may be able to use your GDPR rights to object or request restriction of processing to limit its use for AI training going forward (although platforms may assess these requests on a case-by-case basis).

For your content (including your copyrighted works), there may be AI/content-use settings you can use to restrict the platform’s claimed permissions, so your posts are excluded from future training. Using both may be the strongest practical step to limit use of your content and data for future AI training by these platforms.

It’s important to note that sharing posts on social media sites that refer to or include messaging such as “I do not consent” posts does nothing legally. The only effective route is the official objection process.

Here is guidance on how to opt-out of your personal data being used to train AI models on some of the most used social media and other platforms:

How to opt-out on platforms

Meta platforms 

Meta platforms include Instagram and Facebook. In the UK and EU, Meta provides a “Right to Object” or “Submit an objection request” form in each of its platforms Privacy Centres (you can also find links to these forms under sections on the platforms that are titled “AI at Meta”). 

You must file an objection request or toggle off the AI training button for each account separately, and in the objection request forms you’re also asked to explain how AI training affects you. 

You may wish to explain how the use of your personal data for AI training affects your ability to control your creative works or data protection rights.

How to opt-out:

1. Open Facebook or Instagram.
2. Go to Settings & Privacy → Privacy Centre → AI at Meta.
3. Tap or click Submit an objection request / Right to object.
4. Fill in the form (email + short explanation of why you object to your data being used for AI training).
5. Repeat for each account (Instagram and Facebook use separate flows).
6. If you’re in the UK, EU/EEA or similar jurisdictions, Meta says it will honour these objections. Outside those regions, the same level of objection may not be offered.

LinkedIn

LinkedIn says that it uses some member data to train AI models by default. You can turn this off in Settings. LinkedIn also says you can submit a “Data Processing Objection” (a formal GDPR-style objection) telling them not to use your data for content-generating AI. 

Note: LinkedIn states that opting out mainly stops them using your future data; it doesn’t guarantee removal of data they already ingested. 

How to opt-out:

1. Log in.
2. Go to Settings & Privacy.
3. Go to Data privacy → Data for Generative AI improvement.
4. Turn OFF “Use my data for training content creation AI models.”
5. Then submit a Data Processing Objection (LinkedIn’s GDPR-style objection) to state that you do not want your data used to train generative AI. 

Pinterest

Pinterest updated its Privacy Policy in 2025 to say that Pins (including past Pins) may be used to develop generative AI, including UK users. Pinterest has an explicit Generative AI setting for the content you save/upload, including older Pins that UK users can turn this off. 

How to opt-out:

1. Log in.
2. Go to Settings → Privacy and data → Generative AI (name can vary slightly).
3. Turn OFF the option that says Pinterest can use your content to improve / train its generative AI.

If you’re in the UK/EU you may choose to send Pinterest a “right to object” request under GDPR (Article 21). In this request, you may object to the processing of your personal data for AI training. 

It is advisable to keep any commissioned images off public boards unless you have agreed sharing these publicly within your commission contract. 

Please note: Pinterest makes clear this affects future processing of content; it does not guarantee removal of content already used in AI models. 

Remember to review old boards and move commissioned work to private boards!

WeTransfer and similar file-sharing tools

WeTransfer’s updated Terms of Service caused concern in 2025 because they appeared to grant very broad rights over uploaded files, which artists often use to deliver and send full-resolution, unreleased work. WeTransfer has since stated that it does not train AI on users’ files and clarified its wording. 

Currently, there is no “AI training opt-out toggle” in WeTransfer because the company says that it is not using your files to train AI. However, it is important to note that this could change, and users should continue to monitor any updates to the Terms of Service. 

Steps you can take to help protect your copyright when using WeTransfer:

• When sharing creative work, use password-protected or limited-time transfers.
• Keep a copy of WeTransfer’s current Terms of Service and privacy note in your files to make sure that you have a record to what you originally agreed to when using the service.  
• If you have concerns, you may contact WeTransfer’s data protection or privacy team (privacy@wetransfer.com) to ask how your data or content is used and convey any objections you have to use of your uploads AI or machine learning purposes. 

Key things to look for in any Terms & Privacy Policy

When you read a site’s Terms of Service or Privacy Policy, treat the following phrases as yellow/red flags:

• “To operate, develop and improve the services”
• “For research and development purposes”
• “To develop new features or technologies”
• “We may analyse content using machine learning to improve accuracy and performance”

These phrases are often used to justify using anything that you upload or share to train AI systems, even if they never say, “we train AI on your work.” 

As many social media and other platforms move to relying on default AI training on data and content, often based on “legitimate interest,” not explicit consent. Making sure that you opt-out and/or object to your data being used for AI training is increasingly important. Because platforms are asserting AI training rights through their terms, objecting and turning off default AI training on sites and platforms you use, may be the only practical way to retain some control over how your copyrighted works and content are used by these platforms when it comes to AI training. 

Treating wording with no clear limits, as a red flag

In addition to the consent problem, there is also a transparency problem. Terms now use phrases like “to operate, develop and improve the services” or “for research and development,” which are broad enough to include machine learning and AI training, even if “AI” is not said outright. When you see that kind of wording with no clear limits, you should treat it as a red flag.

Your creative work and your personal data are governed by different rules. When considering AI training, it can help to think about it in the same way as other online rights issues e.g.  – reviewing platform settings, checking whether GDPR rights such as the ability to object may apply, using any available AI/content-use controls, and avoiding posting commissioned or unreleased work publicly. 

Finally, many users keep watch on any platform policy changes and adjust their settings or practices as needed to stay aware of how their personal data and copyrighted content may be used. 

More information

• What is visual arts copyright?
• How DACS is supporting artists and their beneficiaries as AI develops