"Made with AI" label signals

Remove AI label signals from images

Platforms like Instagram and Facebook often show "Made with AI" labels when they detect specific metadata. RAIW removes common provenance markers and metadata that trigger these disclosures.

Why platforms show an AI label

Social media platforms use automated systems to detect generated images by reading embedded metadata fields. The two most common triggers are:

  • C2PA Content Credentials - provenance manifests embedded by generators
  • XMP DigitalSourceType - a metadata field specifically indicating algorithmic media

When either signal is present, platforms often apply the label automatically.

How RAIW reduces these signals

Our tool targets the metadata container. It strips both C2PA Content Credentials and XMP DigitalSourceType values from your image file. It also removes other AI-related metadata like EXIF AI tags.

By removing these embedded fields, you significantly reduce the signals that platforms rely on for automated labeling, all without changing your image's visual content.

Processing and privacy

Signal reduction is achieved by stripping file metadata:

  • Fully secure processing: Metadata stripping happens immediately on our backend.
  • Data handling: We may keep uploaded files and results for up to 30 days to run and improve the service, then delete them; we never sell them.
  • Free: Removing label signals via metadata is completely free.

Frequently asked questions

What signals are targeted by the tool?
The tool primarily aims at C2PA Content Credentials and XMP DigitalSourceType metadata. These are currently the most common signals platforms use to trigger automated AI labeling.
Does RAIW handle the newer IPTC 2025.1 AI-disclosure fields?
Yes. Newer images can carry IPTC 2025.1 AI-disclosure fields (such as the named AI system that made or edited them) alongside the older C2PA and DigitalSourceType tags. RAIW strips these AI-provenance metadata fields too, so they no longer flag the image.
Are labels guaranteed to disappear?
No. Platform detection algorithms constantly update and they may use broader signals, context, or even invisible pixel-level watermarks. We remove the most common proven metadata triggers, but cannot universally guarantee how a third-party platform will evaluate the image.
Do platforms use more than just metadata to detect AI?
Yes. While metadata is the primary method, some platforms are beginning to evaluate hidden pixel markers (like SynthID) or visual anomalies.
What exactly can RAIW influence?
RAIW can cleanly strip the embedded metadata containers and, using the All-in-one mode, disrupt known hidden frequency patterns in the pixels.
Clean AI label signals

No signup required. Visible and metadata removal always free. GPU modes cost $1.25 per image. Supports PNG, JPG, and WebP.