As of early 2026, Midjourney does not embed C2PA Content Credentials — images downloaded from Discord or the web app carry no cryptographic manifest. Instead, Midjourney writes your full prompt text in plain readable form into the image's Description/EXIF field. Anyone who opens the file's Properties panel or an EXIF viewer can read exactly what you typed, including any private or sensitive details in the prompt itself.
Most of the AI-detection conversation focuses on C2PA manifests (used by DALL·E, Adobe Firefly, and others). Midjourney sidesteps that system entirely, which means platform "AI info" labels that specifically look for C2PA or IPTC DigitalSourceType signals may not catch a Midjourney image the same way. But the trade-off is that Midjourney's own metadata is arguably more exposing in a different sense: it's your literal prompt, in plain text, not an encoded technical flag.
username_your_prompt_text_hash.png, which exposes both your Discord username and part of your prompt in the filename itself, independent of any embedded metadata.If your prompt referenced something private — a client's brief, an unreleased product name, a person's name, a location — that text travels with the file wherever it's shared, embedded in a way most people never think to check. Stripping it before publishing or sending the file removes that trail along with everything else.
Because Midjourney's metadata lives in standard EXIF/Description fields, not a C2PA manifest, AiEraser's Remove mode clears it the same way it clears any EXIF data — by re-encoding the image through canvas, which keeps the pixels but drops every text field, entirely in your browser. Use View mode first if you want to see exactly what prompt text and filename metadata is currently embedded.
Sources: independent C2PA-adoption tracking (c2paviewer.com, fast.io, early 2026), community documentation of Midjourney's EXIF behavior. Tool behavior can change between versions — verify against a current Midjourney export if this is decision-critical.