Meta AI Image Generator Sparks Controversy Over Default Privacy Settings

Meta recently launched an AI image generator named Muse, targeting advertising creative, interior design, and creator content production scenarios. The tool is based on an upgraded Emu model and incorporates a style anchor mechanism, supporting the binding of brand elements for image generation. However, the accompanying default data usage settings quickly drew user and regulatory attention, leading to pressure for the feature to be taken down.

Fact Restoration

According to public reports, Muse is deeply integrated into Instagram and Facebook platforms, allowing users to generate Reels covers or background images through natural language instructions. Training data sources include the authorized Shutterstock library and works from partner artists, but there is also a mechanism that defaults to scraping users' public Instagram photos for AI training. Unless users manually disable the setting, their photos may be used for model optimization. This setting has been reinstated in North America, while it was previously suspended in the UK and Europe for similar reasons.

Mechanism Breakdown

Why did this controversy occur? The design intent of Muse is to improve generation quality through large-scale real photos, especially in terms of spatial understanding and brand consistency. The default mechanism allowing the use of public photos effectively leverages user inertia—users who do not actively take action will have their data collected. To prevent future photos from being used, users must navigate to Instagram Settings & Privacy, find the account center, locate the AI training data usage option, and turn off the switch. Whether historical photos have already been included in training, and whether Meta provides a batch deletion option, remains unclear. This "opt-out" rather than "opt-in" default path potentially conflicts with the explicit consent principle emphasized by GDPR, and has been labeled by critics as dark patterns.

Industry Impact

In terms of competitive landscape, Muse attempts to differentiate itself from DALL-E and Midjourney through commercial application scenarios, but the privacy controversy may undermine trust among advertisers. Developers need to reassess the compliance of data sources and shift toward more transparent licensed libraries. When integrating similar tools, enterprise users will face additional review processes to ensure compliance with the human-in-the-loop principle, meaning AI-generated content must be reviewed by humans before publication.

Strategic Judgment (Analysis)

The most likely next step is that Meta will further adjust Muse's data collection strategy, adding transparency prompts or regional restrictions to address regulatory pressure. Similar features on other platforms may also face the same scrutiny, and users proactively checking social media platform privacy options will become the norm. The balance between AI training's reliance on real data and privacy protection still needs to be gradually resolved through clearer mechanisms.