Calabi Labs · Guide · 2026-06-14
Meta will add made with ai labels on images and videos next month pe
Meta's AI detection is getting sharper — and it's looking at your file's metadata, not just what the image looks like. If you've been using AI video or image generators, Meta will start slapping "Made with AI" labels on your uploads starting next month. Here's what that means and how to fix it before it happens to your content.
What Actually Gets Your Content Labeled "Made with AI"
Meta isn't analyzing your pixels to guess if something is AI-generated. It's reading the invisible metadata layer baked into every file — the same way forensic tools work. When you export from Midjourney, Sora, Runway, or Leonardo AI, your file carries a specific signature that automated systems flag immediately.
The primary trigger is C2PA / Content Credentials — a cryptographic manifest embedded in the file that explicitly states the tool used to create it. This lives inside JUMBF boxes in the file structure. Export an AI image from Adobe Firefly or Photoshop's AI features and you'll find the DigitalSourceType: trainedAlgorithmicMedia XMP tag sitting right there in the metadata. Generator-specific tags, encoder fingerprints like Lavc (Lavender encoder, common in AI video pipelines) and x264 SEI messages in video bitstreams — all of these light up Meta's classifiers.
A raw AI export typically carries 140+ metadata tags. Platforms like Instagram and Facebook scan for a specific subset: missing GPS coordinates, a capture timestamp that doesn't exist or shows a future date, an encoder that isn't a known phone model, and the absence of the EXIF fields a real camera would embed. Hit enough of these signals and your Reel or post gets the "Made with AI" label — automatically, within seconds of upload.
Why Cropping, Screenshots, and Re-Uploads Don't Work
Most creators try the obvious fixes first. You crop out the corner of a generated image. You screenshot your AI video and post that. You download, re-upload, maybe compress it through a converter. None of this reliably removes the detection layer because it lives in the file structure itself — not in the visible pixels.
Cropping changes what you see but leaves the JUMBF boxes and XMP metadata intact. Screenshots create a new file, but if you screen-capped an AI export, the screenshot inherits the same problem. Re-encoding disrupts some visible artifacts, but the metadata signals — the C2PA manifest, the encoder fingerprints, the missing device identity — survive. Meta's systems don't need a visible watermark to flag you. They're reading the file's ID card.
The metadata layer is also what survives downloading and re-uploading on different platforms. If you posted an AI video to Reddit and got flagged there, that same file uploaded to Instagram carries the same forensic trail.
How to Actually Clean Your AI-Generated Files Before Posting
Calabi runs a one-pass pipeline that strips the detection signals and injects authentic phone-capture identity — so your file reads as a normal phone recording at the file level. Here's how it works:
- Upload your AI-generated image or video. Drag and drop or select the file. No manual settings, no selecting regions or adjusting parameters.
- Automatic strip. The pipeline removes every signal platforms scan for: all JUMBF / C2PA atoms (reduced from 18 to 0), all C2PA references (16 to 0), the trainedAlgorithmicMedia XMP flag, generator-specific tags, and encoder fingerprints like Lavc and x264 SEI messages. Your 140+ tag AI export becomes roughly 94 neutral structural tags.
- Automatic inject. Calabi writes in authentic phone-capture identity: a real device profile — iPhone 15 Pro, Pixel 8 Pro, or Galaxy S24 Ultra — with Make, Model, Software version, GPS coordinates, and a capture timestamp that matches real-world conditions.
- Review the forensic proof card. Before download, you see the same ExifTool scan platforms use — showing exactly what was stripped and what was injected. This is what gets checked when your file hits Instagram's servers.
- Download the cleaned file. Post it directly. It reads as a phone recording, not an AI export.
Visible watermarks — like Sora's sparkle icon or a platform's corner logo — require cropping to remove because they're baked into the visible image. Calabi removes the invisible detection layer that survives cropping, which is what actually gets you flagged after you've already trimmed the visible mark.
Frequently Asked Questions
Will Meta still label my post if I use Calabi?
Calabi removes the metadata signals Meta's systems scan for. Results vary by platform and source model — no tool can guarantee a platform won't flag you — but Calabi eliminates the specific detection layer that triggers automatic "Made with AI" labels on Instagram and Facebook.
Does Calabi change how my image or video looks?
No. Calabi works entirely on invisible file metadata and structural signals. Your pixels, resolution, color grading, and visual content are untouched. It doesn't edit, paint over, or reconstruct any part of the visible image.
What device profiles does Calabi inject?
Current profiles include iPhone 15 Pro, iPhone 16 Pro, Pixel 8 Pro, and Galaxy S24 Ultra — widely-used devices that write consistent, verifiable EXIF and XMP data. The injected identity is what shows up in the forensic proof card.
Meta's automatic labeling is rolling out to more creators next month. If you're posting AI-generated content on Instagram, TikTok, YouTube, or Reddit, the metadata trail is what determines whether your file gets flagged — not the visual quality. Clean the file before you post, not after.
Try Calabi free at calabilabs.com — 10 cleans, no card.