Skip to main content
Tech Provider forFree foreverNo credit cardUnlimited DMsNo watermark
Compliance

AI Creator Label

Also known as: Made with AI, AI Info

Instagram's opt-in label (launched May 4, 2026) that marks an account as primarily AI-generated content. Auto-detection covers C2PA metadata, watermarks, and audio fingerprinting.

Updated Jun 1, 2026

The AI Creator Label is Instagram's account-level disclosure tag for accounts whose content is primarily AI-generated. Meta launched it on May 4, 2026 as part of a broader content-provenance rollout that aligned with the EU AI Act's Article 50 transparency obligations coming into force August 2, 2026. The label appears under the account name and inside the bio area; tapping it opens an explainer drawer naming the detection signal (creator-declared, C2PA metadata, SynthID watermark, or audio fingerprint).

The label can be applied two ways:

  • Self-declared. Account holder toggles "AI content account" in Settings. Recommended for any creator running a faceless AI-driven channel — voluntary disclosure carries no distribution penalty.
  • Auto-detected. Instagram's detection stack inspects uploads for C2PA Content Credentials (the cryptographically signed provenance metadata OpenAI, Google, Adobe, and Midjourney all began embedding by mid-2026), Google DeepMind's SynthID invisible watermark on images and audio, and Meta's own audio fingerprint database for AI-cloned voices. If detection fires on enough posts, the account is auto-labelled.

Why it matters for creators

The label isn't a punishment — Meta has been explicit that AI-content accounts are not penalised in distribution simply for being AI-driven. What does get penalised: undisclosed AI content that the detection stack later catches. If you run a faceless AI channel and skip the self-declaration, an auto-applied label after the fact can spook brands mid-deal and trigger a temporary reach correction while the algorithm reclassifies the account.

What creators get wrong

  • Assuming detection can be defeated by re-encoding. C2PA metadata strips on most re-uploads; SynthID does not. The watermark survives compression, screenshot, and crop.
  • Thinking it only applies to images. SynthID-Audio detects AI-cloned voiceovers, which is the most common evasion vector on faceless Reels.
  • Conflating the account label with the post label. Instagram has a separate post-level "AI Info" tag for individual pieces of content. The Creator Label is account-level — it's a stronger signal applied to creators whose feed is mostly synthetic.

The label also sits inside the EU AI Act Article 50 compliance chain — any creator reaching European audiences with synthetic content needs to disclose anyway. Self-labelling now is the cheap version of that compliance. Full walkthrough: the Instagram AI Creator Label explainer.

Example

Example. A faceless meditation channel running 90% AI-generated voiceover Reels skips self-declaration through April 2026. On May 14, ten days after launch, SynthID-Audio flags eight of the last twenty Reels and Instagram auto-applies the AI Creator Label. The account's reach drops 22% for six days while the algorithm reclassifies it, then recovers. A neighbouring channel that self-declared on May 4 sees no reach change at all — same content, opposite outcome. The cost of evasion was the brief throttle, not the label itself.

Related terms

Read more