AI labels by platform: what you must label, and what you don't

Short answer: label realistic AI images, video and audio where a platform asks you to. Text you wrote with AI help, such as captions, scripts and replies, isn't covered by the label rules on the networks we checked.

Instagram, Facebook and Threads (Meta)

• Meta's "AI info" label covers images, video and audio, not text (Meta).
• You're asked to use the disclosure tool when you post realistic video or audio made or changed with AI, and Meta may apply penalties if you don't (Meta).
• Content that was only edited with AI shows the label in the post menu, not on the post itself.

TikTok

• Label AI-generated or heavily edited content that shows realistic-looking scenes or people. Unlabelled content may be removed, restricted or labelled for you (TikTok).
• Labels TikTok adds automatically, for example from its own AI effects, can't be removed.

YouTube

• Disclose realistic altered or synthetic content, such as a real person appearing to say something they didn't (YouTube).
• Using AI for scripts, outlines, titles, thumbnails, captions or ideas needs no disclosure.
• Disclosing won't by itself limit a video's audience or its eligibility to earn money.

LinkedIn and X

• We found no official rule on labelling AI-assisted text on either network.
• X labels misleading manipulated media, which is about deceptive edits, not writing help (X).

A quick decision guide

• AI wrote or polished your caption or reply: no label needed on the platforms above.
• AI made a realistic photo, video or voice: label it where the platform asks.
• AI made an obviously stylised illustration: check the platform. Meta and TikTok focus on realistic content.
• You're not sure: label it. YouTube says disclosure alone doesn't cut reach.

Related: Does using AI hurt your reach? · Will customers know it's AI?