Instagram’s New AI Can Turn Your Photos Into Deepfakes. Here’s How to Opt Out

Instagram has quietly launched an AI image generator that can take any photo you’ve shared—or that others have shared of you—and rework it into new, AI-created images. The feature, powered by Meta’s latest generative models, is rolling out to users worldwide. Privacy experts are raising red flags, and for good reason: unless you actively opt out, your public photos may be used to train the tool or manipulated by anyone who has access to it.

What Happened

In early July 2026, news broke that Instagram was testing an AI feature that lets users “reimagine” existing photos in different styles or with altered content. The Guardian reported on July 9 that the feature uses a diffusion model trained on massive datasets, including user uploads from the platform. Yahoo Finance UK published a guide on July 10 explaining how to opt out, noting that most public accounts are automatically enrolled.

Meta has not publicly confirmed the full scope of data used, but multiple sources—including inkl and Yahoo Finance UK—state that the opt-out is only available for users whose accounts are set to public. Private accounts appear less exposed, but the details remain unclear.

Why It Matters

The core concern is consent. Your photos, including ones you did not take or upload yourself, can be fed into an AI model without your explicit permission. This raises risks:

  • Deepfakes. Anyone could take a picture of you and generate a new image that makes you look like you said or did something you never did. While Instagram likely applies some guardrails, the technology is advancing rapidly, and misuse is hard to prevent.
  • Loss of control. Once an image is fed into the AI, there is no guarantee it will not be stored or used elsewhere. Meta has faced previous controversies over data handling, and trust remains low.
  • Viral manipulation. The feature could be used to create convincing fake images that spread quickly, damaging reputations or causing harassment.

The Guardian has also covered the broader threat of AI-generated blackmail, noting that UK schools have been advised to remove pupils’ online photos. This development fits into a worrying pattern.

What Readers Can Do

The good news is you can limit how your images are used. Here is a step-by-step guide to opting out, based on the information available as of July 2026. Note that Instagram may update its settings, so these steps could change.

  1. Open the Instagram app and go to your profile.
  2. Tap the three horizontal lines (menu) in the top right.
  3. Select Settings and privacy.
  4. Scroll to Privacy and then look for a section labelled Data sharing or AI features. (The exact label may vary; some users see “AI model training” or “Generative AI.”)
  5. Find the toggle for “Allow your photos to be used for AI training” or something similar. Turn it off.
  6. If you see an option to “Delete previous training data” – which some reports mention – consider activating it. This may remove your past images from future model versions.

Important caveats:

  • The opt-out appears to work only for accounts that are public. If your account is private, your images are already less exposed, but Instagram may still process them for internal features.
  • Opting out may not affect images already processed before you changed the setting.
  • If someone else posts a photo of you, their settings govern that image, not yours.

Beyond the toggle, take these steps to reduce risk:

  • Switch your account to private. This limits who can see your photos and may block the AI from accessing them.
  • Review tags and mentions. Untag yourself from any photos you do not want used.
  • Avoid posting sensitive images. Anything you upload could eventually be fed into an AI model.
  • Check your privacy settings regularly. Meta often changes menus.

Broader Context

Instagram is not alone. Other platforms owned by Meta, such as Facebook and Threads, have similar features or are testing them. Google, Snapchat, and TikTok have also deployed AI image tools that rely on user data. The privacy debate around AI training data is ongoing, with regulators in Europe and the UK scrutinising these practices.

The Yahoo Finance UK guide points out that while the opt-out exists, it is not prominently advertised—a pattern common in the industry. Users must actively seek it out.

Conclusion

Instagram’s AI image generator is a powerful tool, but its rollout has blurred the line between user content and training data. If you value your privacy, take the few minutes to disable the feature and secure your account. The settings are there, but only if you know where to look.

Sources:

  • The Guardian, “Instagram’s AI image generator alarms privacy experts,” July 9, 2026.
  • Yahoo Finance UK, “Instagram has just allowed anyone to alter your images with AI. Here’s how to opt out,” July 10, 2026.
  • inkl, “Meta’s New AI Can Turn Instagram Photos Into Deepfakes. Most Public Users Are Included Unless They Opt Out,” July 9, 2026.