Meta has introduced a new AI-powered tool called Muse Image, part of its Meta Superintelligence Labs, that can generate images based on user prompts. The tool also allows users to reuse public Instagram posts and Reels—including those from other accounts—to create AI-generated images. This raises significant privacy and control concerns for Instagram users with public profiles.

What is Muse Image?

Muse Image is Meta's latest image generation model, designed to understand complex prompts and blend multiple photos into single images. It can create images from text descriptions or from existing photos uploaded by the user. The tool is free and accessible via Meta AI website, mobile apps, and chats on Instagram and WhatsApp, with Facebook and Messenger support coming soon.

Users can explore suggested prompts (presets) and share generated images to Instagram Stories, feed, or chats. The release includes over 30 new AI-powered effects for Instagram Stories.

How does it use Instagram posts?

The key privacy concern is that Muse Image can reuse any public Instagram post or Reel as a source for generating new images. To do so, a user would download or screenshot a public post, upload it to Muse Image, and describe the desired output. Meta AI then creates a new image based on that content. This means anyone with a public Instagram account could have their photos used without explicit permission.

While Meta frames this as a creative tool—allowing users to "remix" their own photos in fun ways—the feature opens the door to misuse. For example, someone could take a public portrait and generate an unflattering or inappropriate version. While the tool does not directly integrate with Instagram accounts (users must manually capture content), the potential for widespread reuse exists.

Privacy implications and background

The rise of generative AI has repeatedly collided with privacy concerns. Platforms like OpenAI's DALL-E, Midjourney, and Google's Imagen have faced criticism over training data sourced from the web without creators' consent. Meta's approach is similar but more direct: instead of training the model on public data in aggregate, Muse Image actively lets users pull individual posts in real time.

This creates a new vector for identity manipulation. A public Instagram account—often used by influencers, businesses, and celebrities—could be exploited to create deepfakes, misleading images, or parody content. Even if the original post is removed, the AI-generated derivative may still circulate.

Meta's terms of service likely grant the company broad rights to use public content for AI development, as seen in other controversies. However, the company has provided an opt-out mechanism, which is a step toward user control.

How to opt out

Fortunately, disabling this access is straightforward. Follow these steps to prevent your Instagram posts and Reels from being used by Muse Image:

Check your account privacy status

Open the Instagram app and tap your profile icon at the bottom right. Tap the three-line Settings icon at the top. Scroll down to "Who can see your account" and note whether Account privacy is set to Public or Private. If your account is public, the following steps are especially important.

Disable AI-based access

While still in Settings, scroll further down to "How others can interact with you" and select "Sharing and reuse." Look for the section "Allow people to reuse your content on Instagram and with AI features at Meta." Turn off the switches for both Posts and Reels.

Next, under "Allow people to create with and reuse your original audio on Meta AI," turn off the switch for Reels. This prevents your original audio tracks from being used as input for AI generation.

Once these settings are disabled, your public content cannot be used as a source for Muse Image. Your account remains public, but the specific reuse permission is revoked.

Broader context of AI and social media

Meta is not alone in integrating AI with social media content. Snapchat has long used AI for lenses and filters, and TikTok offers AI-generated effects. However, the ability to pull arbitrary public posts as base images is relatively novel. It reflects a trend where user-generated content feeds AI models without explicit consent.

Regulators in Europe and elsewhere are scrutinizing such practices under GDPR and other privacy laws. The opt-out provided by Meta may not satisfy all legal requirements, as it places the burden on users to proactively disable a feature they might not know exists.

For creators who rely on public visibility, this feature could undermine control over their digital likeness. The recommended best practice is to review privacy settings regularly and consider maintaining a private account for personal posts while using a separate public account for professional content.

Technical overview of Muse Image

Muse Image is built on Meta's research in large-scale transformer models. It processes text prompts and image inputs through a diffusion-based architecture, generating high-resolution outputs. The model can handle complex requests such as "create an image of a cat wearing a top hat in the style of Van Gogh" or "erase the person in the background of this photo." It also supports multi-image blending, where multiple source photos are combined into a coherent scene.

Meta claims that Muse Image uses advanced reasoning to understand spatial relationships, object attributes, and stylistic cues. The company has been investing heavily in generative AI, with CEO Mark Zuckerberg announcing plans to build "artificial general intelligence" and integrate AI across all Meta platforms. The launch of Muse Image is a step toward that vision.

But the integration with Instagram posts raises questions about data provenance and ownership. When a user uploads a screenshot of a public post, the AI does not know whether the user has rights to that image. It simply generates a derivative work. This could lead to copyright and trademark disputes, especially if the generated image is used commercially.

Meta's terms for Muse Image likely state that users are responsible for the content they upload. Yet the company also benefits from the vast library of public Instagram content that can be used to improve its models. The opt-out mechanism only prevents real-time reuse, not training on historical data.

Industry reactions and recommendations

Privacy advocates have expressed concern that the feature normalizes the use of personal images without consent. The Electronic Frontier Foundation (EFF) has previously warned about similar features on platforms like LinkedIn and Facebook. They recommend that social media companies should require explicit opt-in for any AI reuse, not just provide a buried opt-out setting.

Some users have already reported that their posts were used to generate silly or embarrassing images by friends. While harmless in many cases, the potential for harassment or identity theft is real. For example, a public figure's photo could be transformed into a fake image of them engaging in illegal activity.

To protect yourself, in addition to the opt-out steps, consider watermarking your images or adding metadata that restricts AI use. While not foolproof, these measures can deter casual misuse.

Meta has stated that it will continue to refine Muse Image based on user feedback, possibly adding more granular controls. Until then, the privacy burden falls on individual users.

As AI becomes more embedded in social media, the balance between innovation and privacy will remain contentious. Meta's opt-out feature is a small concession, but it highlights the need for stronger default protections and transparent data practices.

Source: ZDNET News