Topaz Photo AI Metadata: A Different Problem Than Generative Tools

Why this is worth knowing about

If you're restoring historical or genealogical photos — a use case Topaz is explicitly marketed for — losing the original IPTC fields (subject names, places, dates manually added by a previous owner) during processing can mean losing information that was the whole point of digitizing the photo in the first place. This is a documented, long-running issue in Topaz's own community forums, not a rare edge case.

The software-identification angle

Separately, because Topaz does write its own CreatorTool/HistorySoftwareAgent fields into output files, a processed image carries a record that it went through AI-based enhancement — relevant mainly in contexts like photo competitions with AI-editing restrictions, where several competitions now specifically flag Topaz's generative features (like its cloud-based Bloom tool) as grounds for disqualification.

What to do about it

• If you want to preserve original metadata, export as TIFF where possible (reported to retain more fields than other formats) and verify with a metadata viewer afterward rather than assuming it carried through.
• If you want a clean file with no software trail, run the output through AiEraser's Remove mode afterward — it strips everything, including Topaz's own identifying tags, via canvas re-encoding.