Does Removing EXIF Also Remove XMP and IPTC Metadata?

People say “strip the EXIF” the way they say “delete the metadata.” Those phrases are not the same job. EXIF is one standard — usually the camera’s technical and capture record. IPTC and XMP are other standards that editors, DAM systems, and stock sites commonly write into the same JPEG or TIFF. A tool that only empties the EXIF APP1 segment can leave IPTC and XMP fully readable.

After any wipe, confirm the outbound file: how to verify metadata was removed. Provenance-style labels (including AI-related tags) are a related but separate topic: do AI-generated images have hidden metadata that reveals they were AI-made.

Three standards, one file

EXIF (Exchangeable Image File Format) is what most phone and camera pipelines write at capture: make and model, lens and exposure settings, orientation, capture timestamps, and often GPS when location services were on. When someone opens “Details” or “Get Info” on a JPEG, they are often looking primarily at EXIF fields. That visibility is why “EXIF” became shorthand for all photo metadata — and why incomplete tools feel finished after clearing it.

IPTC grew out of news and publishing workflows. It is where captions, headlines, keywords, bylines, copyright notices, and contact or licensing language commonly live. Photoshop, Lightroom, and many stock or agency pipelines write IPTC when you caption a shot or attach rights language. A marketplace or newsroom tool may read IPTC even when a consumer EXIF viewer shows “no GPS.”

XMP (Extensible Metadata Platform) is Adobe’s extensible packet format. Edit history, ratings, hierarchical keywords, and many modern sidecar-style fields travel as XMP inside the file or beside it. XMP can also duplicate values that also exist in EXIF or IPTC — including location — so clearing one copy does not clear the other.

These blocks are not alternate names for the same bytes. They are separate structures that different software adds at different times. The camera may write EXIF at shutter press; Lightroom may add IPTC captions weeks later; an export may embed XMP describing the develop settings. A “strip EXIF” pass that never opens the IPTC or XMP packets leaves that later work intact.

Why the same file can disagree with itself

Because writers update different blocks independently, a single photo can hold conflicting stories. EXIF GPS might reflect where the camera was when the shot was taken. A later XMP or IPTC location field might reflect a manual pin, a stock category, or an older edit that was never reconciled. Author fields can similarly diverge: camera Artist in EXIF, Creator in IPTC, and another byline in XMP.

Viewers then disagree because they prioritize different standards. One app surfaces EXIF first and looks “clean” after an EXIF-only strip. Another app — or a stock ingestion script — prefers IPTC/XMP and still shows the photographer’s name, keywords, or a second GPS. That is not magic recovery of deleted EXIF; it is reading a block the first tool never touched.

This also explains odd verification results. Someone strips with Tool A, checks with an EXIF-only panel, ships the file, and gets a report that “metadata was still there.” Both observations can be true: EXIF was gone; IPTC/XMP were not.

Real-world implication: the casual EXIF check is not enough

The practical gap is everyday, not exotic. You clean a portfolio JPEG with a basic EXIF remover before posting. A client or aggregator that reads IPTC still sees your caption keywords and copyright string. You remove location from a phone share sheet (often EXIF-oriented) and later discover coordinates still present in an XMP packet written by an editor. You “pass” a quick online EXIF viewer while a fuller inspector lists Creator, Description, and History.

None of that requires inventing scary attack scenarios. It is ordinary software writing ordinary fields into ordinary files — and ordinary incomplete removers leaving those fields alone. For privacy, the outbound copy needs a full wipe of common blocks, not a synonym for “GPS off in EXIF only.”

Workflow that holds up: strip with a tool that targets photo metadata broadly, download the cleaned export, then verify with a viewer that shows more than a single EXIF summary — including author, description, GPS, and software or history-style fields. The verification habit is documented in how to verify metadata was removed.

What “remove all metadata” has to mean

“All” in day-to-day photo privacy means: clear the EXIF block, clear IPTC, and clear XMP in the file you will actually attach or publish — then confirm those fields are empty on that export. It does not mean hoping a social app recompresses the file into cleanliness. It does not mean trusting a button labeled only “Remove EXIF” without checking what remains.

Edge cases still exist (format-specific chunks, thumbnails, paired Live Photo resources), but the three-standard gap is the one most people hit first. Treat EXIF-only language as a red flag in any tool description. Prefer a wipe that rewrites a clean outbound image and a second-look verification step over a one-click claim that never mentions IPTC or XMP.

MetadataWipe is built for local, in-browser stripping of common photo metadata so the file never needs to leave your device for the wipe itself. After you download the cleaned copy, re-open it and read the panel — same discipline as any other strip. For unusual provenance labels that are not classic EXIF GPS, see also AI image hidden metadata.

Common mistakes

Equating “no EXIF GPS” with “no metadata.” Bylines, captions, and XMP history can remain.

Verifying only with an EXIF-only viewer. Use a check that can surface IPTC/XMP-style fields too.

Stripping a derivative while sharing the library original. Clean the exact file that will travel.

Assuming one app’s Info pane equals every reader’s view. Different apps prefer different standards.

Stopping at “Remove Location” in a share sheet. That control is not a full three-block wipe.

Related guides

See also:

• How to verify metadata was removed
• Do AI-generated images have hidden metadata that reveals they were AI-made?