What is the "Made with AI" / "AI info" label?
The AI info Instagram label and the "Made with AI" tag are disclosure notices that Instagram, Facebook and their sister platforms attach to a post at upload time. That label can come from the file contents. Platforms scan each upload for provenance markers such as C2PA Content Credentials, XMP packets and IPTC fields like trainedAlgorithmicMedia. When those markers say the file was created or edited with AI, the label goes on. Platforms can also label from their own visual classifiers, which no metadata edit touches. The limitations below spell that out.
The label describes the file itself, whatever your intent was. AI generators write these records at export, and ordinary editing tools do the same. A photo you shot yourself can pick them up after a single round trip through an AI-assisted editor. The pixels stay yours while the paperwork inside the file says otherwise.
Because the trigger lives in the file, the fix lives in the file too. Remove AI label markers before uploading, and the platform has nothing in the metadata to read. This page runs a forensic inspection first. It shows you what your file carries, then strips those markers while keeping your image at its original dimensions and framing.
Why real photos get labeled as AI
Most mislabels come from routine AI-assisted edits. Generative Fill in Photoshop signs the whole export after filling one corner. AI sky replacement marks every file in a product batch. AI Denoise in Lightroom writes an AI edit marker into the output. Upscalers and background removers sign on the way out. A reel cover exported from CapCut carries metadata of its own, separate from the video.
The pattern holds across all of them. One AI step anywhere in the chain leaves a marker in the file, and that marker is what the platform reads. A photographer posting a wedding shot after background cleanup gets labeled. An e-commerce seller who puts a whole catalog through AI background removal sees the batch flagged.
None of those AI-assisted edits change whose photo it is. Stripping the markers before upload removes paperwork that misdescribes the work.
How to turn off AI info on Instagram
There is no setting that turns the label off. Instagram and the other apps build it from the contents of the file at upload time, so no toggle can keep it away while the markers remain inside the file. Searches for how to turn off AI info on Instagram lead to a switch that does not exist.
What the apps do offer is a report path. On some posts you can flag a label you believe is wrong and request a review. A review request acts only after the label has already been applied, and the platform decides the outcome.
Guides that walk you through tapping three dots to edit or remove the label point to an interface that does not exist, or no longer does. Everything that reliably works happens in the file, before you upload it.
How it works
How to remove the AI label before you post
Step 1: Export the final file
Finish your crop, color and captions first, then export once. Every re-export from an editor can write fresh credentials, so cleaning has to be the last step before upload.
Tip
Both Photoshop and Lightroom can export with Content Credentials switched off.
Step 2: Inspect and strip the markers
The forensic pass reads your file and reports the provenance records it finds: C2PA manifests, XMP packets, IPTC source-type fields, EXIF records and PNG text chunks. The removal pass then deletes the whole metadata stack platforms read for labeling, Content Credentials and their JUMBF wrapper included. Your image keeps its dimensions; PNG and WebP files keep their original quality, and a JPEG is saved once more at high quality.
Tip
The inspection shows you what your file actually carries before anything is stripped. Sometimes the honest answer is that there is nothing to remove. SynthID is woven into the pixels and survives any metadata edit, so a Google-generated file needs the pixel layer handled separately.
Step 3: Download the clean copy and post it
The output keeps your image’s dimensions and framing, with the markers gone. The pass also removes the embedded color profile (ICC) with the rest of the standard metadata, so colors display at their standard default values. Upload the downloaded copy; the original you inspected still carries every marker. The strip takes away the metadata trigger, though a platform’s own visual classifier can still label a clean file. Keep the untouched original as your private master, in case you ever need the embedded history back.
Tip
Check the AI-disclosure toggle when you post. If you opted into a disclosure yourself, the platform keeps that record on its side. The strip pass cannot remove it.
Which platforms read AI labels from file metadata
Instagram
Feed posts, stories and reels all show "AI info". When the file is uploaded, Instagram reads C2PA credentials, XMP and IPTC markers, plus AI fields in EXIF.
Facebook
Facebook uses the same "AI info" wording and the same signal set. Both apps are Meta properties.
TikTok
The TikTok label reads "AI-generated" and carries an AIGC tag. It keys off C2PA credentials and IPTC digital-source-type fields.
Pinterest
Pinterest calls the label "AI modified" and applies it from C2PA credentials and AI provenance metadata.
LinkedIn
LinkedIn shows "AI-generated" when C2PA Content Credentials are present on a supported format.
YouTube
YouTube asks creators to declare altered or synthetic content at upload, and can add a label on its own when a file carries C2PA metadata or its systems detect AI. A strip cannot untick a disclosure you already set; that one changes in YouTube Studio.
Meta AI images
Meta AI files carry C2PA credentials and IPTC fields, the markers the AI info label reads on Facebook and Instagram. The visible "Imagined with AI" corner mark and Meta's own invisible watermark sit in the pixels, so a strip leaves both.
X (Twitter)
X builds AI-generated media labels from provenance data and runs visual detection alongside it. Cleaning the metadata removes what X can read from the file; the platform still applies its own judgment.
Which AI tools write these markers
If the post is already up
A label on a published post cannot be edited or removed in place. The platform processes uploads at publish time and never re-scans the live media. No file tool can reach into a live post and rewrite the flag. Anyone promising otherwise is guessing.
The route is manual: delete the post, clean the source file on this page, and repost it. The real cost is engagement, since comments and likes do not transfer. Weigh that against how much the label matters for that particular post.
Before deleting anything, try the report path where the app offers it. If the review succeeds, you keep the post and lose nothing. Clean-then-repost is the fallback.
Stop the markers at the source
The fix at the source is an export setting. Photoshop can leave Content Credentials off entirely: switch it off once in the Content Credentials panel, and exports stop carrying the record. Lightroom has the same switch in its export dialog. One toggle saves a strip pass on every file you publish.
Watch the chain around the export, too. Converters and "save for web" steps can re-embed metadata from the wrong template. The order matters: edit, export once, clean, post. Cleaning in the middle of the chain only gives the next export a chance to re-sign the file.
If you want to check before you post, the inspection pass on this page reads the file and reports what is inside.
Limitations and honest expectations
FAQ
Frequently asked questions
Answers to the questions people ask most about the AI info label and about stripping AI label markers from a file.
Explore more tools
How to Remove AI Info from Instagram Post
The Instagram-specific workflow: what triggers the mislabel, the signals the app reads, and how to repost clean.
How to Remove AI Info from Instagram Story
The story variant: clean the file before you publish to take the metadata trigger off the table.
Metadata Remover
The general metadata pass for EXIF, GPS and editing history, with standard orientation and color fields kept.
AI Metadata Remover
A pass aimed at the AI provenance fields: C2PA credentials, IPTC digital-source-type values and XMP generator tags.
AI Watermark Remover
For marks outside the metadata layer: visible watermarks and pixel-level signals, handled in the same upload.
C2PA Remover
Target Content Credentials directly: inspect the signed provenance stack first, then strip it.
SynthID Remover
The pixel layer a metadata pass cannot reach, for Gemini and Imagen outputs.
Facebook AI Label Remover
Facebook-specific notes on the AI info label: what the upload carries, the report path, and the strip before your next post.
Pinterest AI Label Remover
Pinterest calls its badge AI modified. This page covers the provenance fields it reads and the clean export that goes up without them.
YouTube AI Label Remover
The altered-content disclosure YouTube asks for, how to change it on a published video, and the metadata pass that keeps the next upload clear.
TikTok AI Label Remover
How TikTok builds its AI-generated tag from C2PA and IPTC fields, and what to strip so a fresh upload arrives without the file-side trigger.
X AI Label Remover
X labels from provenance data and its own detection. This page covers the metadata side and is plain about what the platform still decides.
LinkedIn AI Label Remover
LinkedIn reads C2PA credentials on supported formats. Clean those before upload and the file-side trigger is gone; classifiers remain a separate matter.
Meta AI Metadata Remover
The Meta AI case: the C2PA and IPTC records behind the AI info label, plus a straight account of what the strip leaves in place.
All tools
Every remover and detector on one page, grouped by what each one cleans or checks.
Post without the AI label
Inspect what the file carries, strip the markers, and there is nothing in the metadata for the platform to read at upload. Your image keeps its dimensions and framing.