The shared signals
Automatic labeling usually rests on a short list of common signals:
- C2PA Content Credentials declaring AI generation or AI editing.
- The IPTC DigitalSourceType field, such as trainedAlgorithmicMedia.
- Invisible watermarks from some AI providers, where the platform has a matching detector.
- Creator disclosure — a toggle or checkbox when posting realistic AI content.
Meta: Instagram, Facebook, Threads
Meta shows “AI info” (renamed from “Made with AI” in 2024) when it detects industry-standard AI signals or when a creator discloses AI use. After photographers objected to lightly retouched photos being tagged, Meta moved the label for AI-edited content into the post menu and kept the more visible label for substantially generated content.
TikTok
TikTok requires creators to label realistic AI-generated content and has said it automatically labels uploads that arrive with C2PA Content Credentials identifying them as AI-generated.
YouTube
YouTube asks uploaders to disclose realistic altered or synthetic content and shows viewers a label. It also uses C2PA in the other direction, with a “captured with a camera” style indicator for content carrying authentic capture credentials.
Pinterest and LinkedIn
Pinterest has added labels for AI-generated or AI-modified images based on metadata and its own classifiers. LinkedIn shows a Content Credentials icon on images that carry C2PA data, so viewers can inspect source and edits.
Every platform revises these policies often; check the current help center before relying on any particular behavior.
When a real photo is tagged
One generative edit, such as removing a passer-by, can mark the entire photo as composite AI content in its metadata. Before posting, lay the file in AI Label Remover to see which field would trigger a label. If the label would be wrong, washing a copy removes the cause; if the image is substantially generated, disclose it — that is what labels are for.
About AI labels on social platforms
Does Instagram label AI-generated profiles too?
Instagram has introduced profile-level labels for accounts that feature AI-generated people and may limit recommendations for accounts that do not disclose. Creators who only edit with AI tools are generally not affected; check Instagram’s current guidance.
I only removed a small object — why the AI label?
Many editors record any generative edit in C2PA or IPTC as “composite with AI”. Platforms read the declaration without weighing how big the edit was.
Do platforms keep my metadata after posting?
Most read it during upload and strip it from the public copy, so downloading your own post rarely shows the original fields.
Does a washed image avoid every label?
Not always. Washing removes in-file signals, but platforms can also use watermark detection, classifiers or ask you to disclose.