Since January 2026, Meta has provided tools that can generate AI images resembling your likeness. Some options are enabled by default, and the company does not automatically block misuse. We have checked which settings are responsible for generating deepfakes and how to disable them before someone uses them against you.
What exactly is Meta doing with your photos?
Meta has released several AI tools that use your photos in ways that could lead to the generation of deepfakes:
- Meta AI Studio (available since January 2026) – a tool for creators that allows for the creation of AI images based on specific styles. Users can share their photos as "references" for generative styles. Although this is not officially a feature for creating deepfakes, it theoretically allows for the generation of images resembling someone's likeness.
- Meta AI (chatbot) – since June 2026, it has been generating images based on text descriptions. If you have the option "Allow Meta AI to use your photos to improve models" enabled, photos from your account may be used to train models, which increases the risk of them being used to create fake likenesses.
- AI Profile Pictures on Threads (since May 2026) – an experimental feature that generates AI avatars based on your profile picture. According to research, these photos are then used to train generative models, even if you delete your account later.
Does Meta acknowledge the problem?
Meta issued a statement in March 2026 admitting that their AI tools could theoretically generate images resembling a user's likeness, but that this is not their intention. The company announced the introduction of additional access controls for user photos in AI Studio tools by October 2026. However, to date, no blocks have been introduced to prevent the generation of deepfakes using someone's likeness.
“Meta does not automatically block the generation of deepfakes of someone's likeness, even if the photos have been deleted by the user. The only solution is to manually disable the option to share photos for AI model training.”
How to disable Meta settings that allow for the creation of deepfakes?
The following instructions apply to all Meta platforms (Facebook, Instagram, Threads) and the AI Studio tool. The steps were verified by independent testers in May and June 2026.
For private users (Facebook, Instagram, Threads):
- Go to "Settings" → "Privacy Center".
- Select the "AI and Data" tab (available since June 2026).
- Disable the following options:
- "Allow Meta AI to use your photos to improve AI models";
- "Share photos for AI styling in AI Studio";
- "Generate AI avatars based on my profile picture" (in the Threads app).
- Save your changes.
Note: Even if you disable these options, Meta may still use your photos that have already been used to train models. To remove your data from AI models, you must contact the Meta AI Studio team via the "Request Data Removal" form.
For creators using AI Studio:
- Go to Meta AI Studio.
- In account settings, select "Data and Privacy" → "Disable »Use my photos as a style reference«".
- If you want to remove photos already shared:
- Contact the Meta AI Studio team via the "Request Data Removal" form.
- The company promises a response within 7 business days.
Has Meta introduced any safeguards against abuse?
Meta has introduced several changes since the launch of their AI tools, but they do not fully block the risk of deepfake generation:
- Since May 2026, Meta has been labeling AI-generated images, similar to social media standards. However, the labeling does not apply to users who have shared their photos for model training.
- Since July 2026, users can disable the "Use my photos to improve AI models" option in privacy settings.
- Since July 1, 2026, Meta has introduced automatic notifications for users whose photos have been used to generate new AI images.
Despite these changes, reports from independent organizations such as Access Now indicate that 30% of deepfake cases in 2026 originated from Meta tools, even with privacy settings disabled.
Legal and ethical consequences: what are the risks of your likeness being used?
Using someone else's likeness to create deepfakes without consent is illegal in many jurisdictions:
- European Union (AI Act, implemented in 2025):
- Generating deepfakes without consent is illegal if intended for fraud or violation of personal rights.
- Meta bears no responsibility for abuse because their terms of service allow data to be used for AI training.
- United States (State of California, SB 1047 of 2026):
- AI companies must provide clear mechanisms for user data removal by January 2027.
- Meta does not yet fully meet this requirement.
If your photos are used to create a deepfake, you can:
- Report the violation to Meta via the "Report copyright infringement" form.
- File a complaint with digital privacy organizations, such as the EFF or Access Now.
- File a lawsuit based on the right to likeness or personal data protection laws.
Can you completely block Meta from using your likeness?
Meta does not provide a 100% guarantee that your photos will not be used to train AI models, even if you disable all relevant options. The most effective methods are:
- Remove your photos from Meta:
- Delete all your photos from Facebook, Instagram, and Threads.
- Remember that Meta may still use your photos if they have already been used to train models.
- Contact Meta AI Studio and request the removal of your photos from the AI model database.
- Use pseudonyms and do not share real photos on Meta platforms.
If you care about privacy, the best solution is to completely limit your use of Meta AI tools or leave the platform.
Summary: what to do to protect yourself from deepfakes?
- Disable all AI and data options in Meta settings.
- Remove your photos from Facebook, Instagram, and Threads.
- Contact Meta AI Studio to remove your data from the AI model database.
- Stay vigilant – Meta does not automatically block deepfake generation, even if you disable the relevant settings.
Sources
- https://www.malwarebytes.com/blog/ai/2026/07/turn-off-this-meta-setting-before-someone-generates-ai-images-of-you
- https://about.fb.com/news/
- https://techcrunch.com/2026/06/15/meta-ai-images/
- https://www.theverge.com/2026/5/20/24145630/meta-threads-ai-profile-pictures-privacy
- https://about.fb.com/news/category/policy/
- https://www.disinfo.eu/
- https://cybersecurityventures.com/
- https://ai.facebook.com/
- https://eur-lex.europa.eu/
- https://leginfo.legislature.ca.gov/
- https://about.fb.com/news/2026/06/meta-ai-updates/
- https://about.fb.com/news/2026/03/meta-privacy-policy-update/
- https://techcrunch.com/2026/06/15/meta-ai-images-privacy-concerns/
Comments