AI-generated profiles

My role + process

Because the system was built on language, content design led the work. I served as both content designer and product designer. With Policy, Legal, and Integrity, I defined which accounts needed a label, then designed the screens and wrote the copy for a three-step journey:

• Invite creators of AI-generated profiles to label themselves.
• Label those profiles clearly, wherever they appear.
• Enforce the label on unlabeled profiles, with a fair path to appeal.

We launched an optional label first to learn from it, then an enforced version.

Key content design decisions

1. Self-disclosure upsell: Lead with trust, then reach

The system only works if creators label themselves, so I designed the invitation to make disclosure feel like a benefit, not a penalty. The prompt on their profile leads with what labeling does for their audience, building trust by helping people understand what they're seeing, then adds that it can help them avoid limits to their reach in the future. I also made the label easy to find in Edit profile, as a simple toggle alongside fields like pronouns and gender, so disclosure feels like part of describing who you are.

2. Profile label: "AI-generated profile," explained one tap away

I first championed "AI creator" over "AI profile," which implied the account itself was fake. When testing showed viewers mostly wanted to know whether the person was real, I moved to "AI-generated profile," and placed it right under the creator's name, where people decide whether to follow. One tap opens a short explanation that credits the disclosure to the creator, not Instagram, and explains what the label means in plain language.

3. Enforcement: Explain the decision, not just the label

When our systems detected an unlabeled AI profile, the label was applied automatically, so creators needed to understand what happened without feeling punished. The notification tells them in one line what changed and why, using hedged language that reflects detection is an estimate, and links straight to the details. The explanation screen lays out the labeling rules for profiles and content in plain terms, so the decision feels grounded in clear policy rather than arbitrary.

4. Appeals: A fair process creators can trust

Automatic labels only feel fair if creators can push back, so I made requesting a review easy to find and easy to complete. The review sheet starts by admitting that we don't always get it right, then explains what happens next and how long it takes. Once submitted, creators get an instant confirmation, and the page keeps a record of their request, so they're never left guessing.