TL;DR: YouTube re-encodes every video you upload, transcoding it to VP9, AV1, or H.264 and rebuilding the MP4 container from scratch. That process discards almost all of the original file-level metadata: the QuickTime/MP4 udta user-data atoms, the ©xyz GPS location tag, the device-model and software strings, and most XMP packets. What survives is not pulled from the file — it is the data you type into YouTube Studio (title, description, tags, recording date, and any location you set) plus information YouTube derives during processing. So a viewer downloading your video through normal means cannot read your camera's serial number or the GPS coordinates baked in by your phone. The catch: the original file you stored or shared elsewhere still carries everything, and the metadata you enter in Studio is YouTube's to keep.
What metadata does a raw video file contain?
Before a video ever reaches YouTube, the file your camera or phone produced is dense with embedded data. An MP4 or MOV is a container built from nested "atoms" (also called boxes), and metadata lives in several of them. The moov atom holds the structural map of the file, and inside it the udta (user data) atom is where device makers stash descriptive fields. On an iPhone, that includes a ©xyz atom carrying the GPS coordinates where you hit record, a com.apple.quicktime.make and model string identifying the device, the iOS version under a software tag, and a precise creation timestamp.
This is the same class of hidden data that still photos carry as EXIF, just expressed in a different container format. We covered the video-specific version in detail in our piece on what metadata iPhones add to videos, and the GPS exposure is identical in mechanism to the photo case. A 30-second clip filmed in your living room can encode your home's latitude and longitude to within a few meters, the exact phone that shot it, and the second it was captured. None of that is visible when the video plays. It sits in the container header, readable by anyone who opens the raw file with a parser.
Does YouTube strip that metadata when you upload?
Mostly, yes — but as a side effect of how YouTube processes video, not as a deliberate privacy feature. When you upload, YouTube does not store and serve your original file. It runs the upload through a transcoding pipeline that re-encodes the video into multiple resolutions and codecs (VP9 and AV1 for modern clients, H.264 for compatibility) and repackages each rendition into a fresh container. Because the container is rebuilt, the original udta atoms, the ©xyz GPS tag, the device strings, and embedded XMP packets are not carried forward. They were properties of a file that no longer exists on YouTube's side.
We tested this the way anyone can: upload a clip shot on an iPhone with location services on, then use yt-dlp to pull the highest-quality rendition back down and inspect it with a container parser. The returned file's udta atom contains no GPS atom, no Apple device model, and no original capture timestamp. What it does contain is YouTube's own handler metadata and encoder tags — the signatures of Google's transcoder, not your phone. So in the specific sense most people mean, the answer is that YouTube does not pass your camera's embedded location and device data through to viewers.
Photo by Vitaly Gariev on Pexels.
What metadata survives or gets added?
Stripping the file's original metadata is only half the story. YouTube replaces it with metadata of its own, and some of that is information you hand over voluntarily. The title, description, and tags you type into YouTube Studio are stored and, in the case of title and description, shown publicly. The recording date and recording location fields in Studio's advanced settings are optional, but if you fill them in, you have re-attached exactly the kind of when-and-where data the transcode just removed. A creator who strips GPS from their file and then types their home city into the "recording location" box has undone the privacy benefit by hand.
YouTube also derives and retains metadata you never see directly: the upload timestamp, the IP address and account associated with the upload, processing logs, and content-identification fingerprints used by Content ID. This derived layer is governed by Google's data practices rather than anything in your file, and you cannot strip it — it is the cost of using the platform. The YouTube privacy policy and Google's broader privacy documentation describe what gets collected at upload. The honest framing is that YouTube removes the metadata your device embedded and substitutes metadata that Google and you provide.
The original file is still the real exposure
Here is the part that gets lost in the "does YouTube strip it" question: the transcode protects the version on YouTube, not the version on your hard drive. The MP4 sitting in your camera roll, the copy you AirDropped to an editor, the file you uploaded to a client portal or posted to a different platform — all of those still carry the full udta atom, the GPS coordinates, and the device fingerprint. If your threat model is a stranger downloading your public YouTube video, the re-encode covers you. If it is the raw file leaking through any other channel, YouTube's pipeline has done nothing for you.
This matters most for creators who repurpose footage. The same clip often goes to YouTube, gets cross-posted as a short elsewhere, and is handed to collaborators. Each platform handles metadata differently — we mapped the inconsistencies in our look at how TikTok treats video metadata and found that you cannot assume one platform's behavior generalizes to another. The reliable move is to clean the source file once, before it goes anywhere, so that every downstream copy starts from a stripped baseline rather than relying on each service to scrub it for you.
Photo by Jakub Zerdzicki on Pexels.
Does the answer change for YouTube Shorts?
It does not change in the way that matters. YouTube Shorts run through the same transcoding pipeline as long-form uploads, so a vertical clip you post as a Short is re-encoded and repackaged exactly like any other video, and the original udta atoms and GPS tag are dropped in the same step. The difference is in how Shorts get created and sourced. Many creators film Shorts directly in the YouTube mobile app or pull them straight from their camera roll and post in one tap, which makes it easy to skip any cleaning step entirely. The footage is also frequently re-shared across platforms within minutes, so the same raw file lands on several services before you have thought about metadata at all. Because Shorts are short and casual, people apply less scrutiny to them than to a flagship upload — and that is precisely when an unstripped source file with home GPS coordinates quietly spreads to the most places. The platform-level protection is identical; the human workflow around Shorts just creates more chances for the original file to leak before it is ever cleaned.
How to remove metadata before you upload
Cleaning a video before upload is straightforward and worth doing even though YouTube re-encodes. It guarantees the source file is safe regardless of where else it travels, and it removes any chance that a Studio field auto-populates from the file's embedded recording location. Open Metadata Cleaner, drag in the MP4 or MOV, and confirm the strip. The tool rewrites the container without the GPS ©xyz atom, the device make and model strings, the software tag, and the original creation timestamp, then hands back a clean copy. Processing runs locally in the browser, so the file is never sent to a server — which matters when the whole point is to control where the file goes.
After cleaning, upload the stripped copy to YouTube Studio rather than the original, and leave the optional recording-date and recording-location fields blank unless you have a reason to publish them. That combination — a scrubbed source file plus restraint in the metadata fields you fill in — gives you the strongest position. The same workflow applies to footage from any source, including drone and mirrorless cameras, which we covered for the delivery case in our guide on whether videographers should remove metadata.
What stripping metadata does not do
We try to be precise about limits, because overpromising on privacy is its own harm. Removing container metadata does not make a video anonymous. It does nothing to the visible content of the frame — a street sign, a house number, a reflection, or a recognizable skyline geolocates a clip just as effectively as a GPS atom, and no metadata tool touches pixels. It does not affect the upload-side data YouTube collects: your account, IP, and timestamp are recorded regardless of how clean the file is. It does not remove Content ID fingerprints or audio watermarks that a rights holder may have embedded in the soundtrack. And it does not retroactively clean copies of the file you already shared.
What metadata removal does is close one specific, well-understood leak: the structured, machine-readable location and device data that rides inside the file and that automated tools can extract in bulk without ever watching the video. That leak is real, it is invisible, and it is the one you can actually control. Treat stripping as a baseline hygiene step, not a cloak.
Try Metadata Cleaner free
If you publish video anywhere, strip the source file before it leaves your device. Try Metadata Cleaner free — it runs entirely in your browser, removes GPS and device metadata from MP4 and MOV files in seconds, and never uploads your footage to a server.