Meta Retired Its New AI Image Model in Days—Here’s How to Opt Out of Training Data

Last week, Meta launched Muse Image, its first in-house AI image generator. By the end of the week, it was gone. The model, which turned Facebook and Instagram posts into new images, was retired on July 13—just five days after its July 8 debut. While Meta hasn’t released a detailed explanation, reports suggest that user backlash over how the model used personal content to train itself forced the company to pull the plug. For anyone worried about their photos, posts, or conversations being fed into Meta’s next AI system, this is a clear warning sign. Here’s what happened, why it matters, and—most importantly—how to opt out before the next model arrives.

What Happened

On July 8, 2026, Meta announced Muse Image, an AI model that could generate images based on text prompts, similar to services like DALL-E and Midjourney. Unlike those tools, Muse was trained partly on user content from Facebook and Instagram, including public photos, captions, and engagement data. By July 13, the feature was disabled. According to Technology Org, the rapid reversal was driven by an “opt-out speedrun”—a wave of users discovering that their content had been used without explicit permission and quickly adjusting their privacy settings.

Meta’s official statement was brief, citing “feedback from our community” as the reason for the retirement. It’s not clear if regulatory pressure from Europe or the United States played a role, but the timeline is unusually short for a major product launch. Whether Muse Image returns in a modified form remains unconfirmed.

Why This Matters for Your Privacy

The Muse Image episode is a signal that Meta is serious about building generative AI into its platforms, but also that it is sensitive to public pushback. The model’s retirement shows that user pressure can work—but only if you know what settings to change. The underlying infrastructure for collecting data remains in place. Meta’s privacy policy already allows the company to use your content to train AI models unless you explicitly object. If you didn’t opt out before Muse was launched, there’s a decent chance your public photos and posts were used.

Beyond the immediate case, this sets a precedent: with each new AI tool, the window to protect your data may be narrower. The faster Meta cycles through models, the harder it is to keep up. Knowing how to adjust your settings now gives you a head start.

What Readers Can Do: Opt Out of Meta’s AI Training

The following steps work on both the Facebook and Instagram apps (iOS/Android) and their web versions as of mid-2026. The exact path may change, so check periodically.

Step 1: Open Your Privacy Settings

  • On the Facebook or Instagram app: tap your profile picture → Settings & Privacy → Privacy Center.
  • On the web: visit facebook.com/settings or instagram.com/accounts/privacy_and_security.

Step 2: Locate the Generative AI Controls

Inside the Privacy Center, look for a section labeled Data Use or Generative AI. This may appear under “Your Information” or “AI & Machine Learning.” (The label varies by platform and region.)

Step 3: Toggle Off Data Sharing for AI Training

You’ll see a setting like “Allow Meta to use your posts, photos, and interactions to train AI models.” It may be on by default. Switch it off.

  • Note: This setting applies only to future content. Data already collected before you opt out may still have been used.
  • On Instagram, the same control may be under Account Privacy > Data Sharing.

Step 4: Check Third-Party Activity

Even if you opt out of direct training, Meta can still collect data from your activity outside its platforms (e.g., websites with embedded Like buttons). To limit this:

  • Go to Settings → Your Information → Activity Off Platform.
  • Disconnect any apps or sites you don’t want Meta tracking.

Step 5: Review Whether Your Data Was Already Used

Meta does not offer a simple “check if my data was used” tool for specific models like Muse. However, you can:

  • Use the Download Your Information feature (in Settings → Your Information) to see a copy of what Meta stores.
  • For a rough idea, look for any new AI-generated images in your feed that resemble your own posts—Meta sometimes tags generated content with small labels (e.g., “Created with AI”).
  • Third-party services like the Algorithm Transparency Initiative offer audits for some platforms, but they are not official and may not cover the latest models.

If you live in the European Union or the UK, you have additional rights under GDPR to request information about how your data was used. File a data subject access request through Meta’s dedicated portal.

What This Signals for the Future

The quick retirement of Muse Image suggests that Meta is testing the limits of what users will accept. Each launch and pullback refines their approach—and could lead to models that are trained on opt-in data only, or to more aggressive attempts to collect consent through obscure terms-of-service updates.

Staying protected means treating your privacy settings as a recurring chore, not a one-time fix. Check them every few months, especially before major product announcements. Also consider adjusting your post visibility to “Friends” instead of “Public,” because public content is far more likely to be scraped.

Sources

  • Technology Org, “Opt-Out Speedrun: Meta Retires Muse Image Days After Launch,” July 13, 2026.
  • Technology Org, “Meta Launches Muse Image, Its First In-House AI Image Model,” July 8, 2026.
  • Meta Privacy Policy (current version, as of July 2026).

Note: The exact reason for Muse’s retirement has not been independently confirmed. The steps above were tested on standard accounts in the United States; settings may differ by region.