# Remove metadata from a PowerPoint presentation

> A PowerPoint file keeps its author and last editor, company and manager, review comments with each commenter's name, a preview image of the first slide, and every photo you dropped onto a slide with that photo's own EXIF data, GPS included. This tool removes those from .pptx files, along with co-authoring records and add-in tags, and then lists what a second pass still finds in the cleaned deck. Speaker notes are part of the presentation's content, so they are kept, and the result says so.

## Why decks leak more than you expect

Presentations are assembled from pieces. A team member pastes in a chart, someone else adds photos from a site visit, a manager leaves review comments, and the deck goes out to a client or a conference. Along the way, the file gathers names and history:

- **Properties**: author, last saved by, title, subject, created and modified dates, revision count, company, manager and total editing time.
- **Comments**: review threads with each commenter's name. Newer versions keep a separate list of comment authors.
- **Photos on slides**: when you insert a photo, the image file inside the deck can still carry its EXIF block, including GPS coordinates and the phone model. Nobody sees this in slideshow mode, but anyone who unzips the .pptx can read it.
- **Preview thumbnail**: a small image of the first slide, saved so file browsers can show the deck without opening it.
- **Custom properties and custom XML** added by templates or document-management systems.
- **Co-authoring records**: decks edited by several people at once keep a log of who changed what, with account names and times.
- **Add-in tags**: small name-and-value notes that add-ins attach to the deck or to single shapes. An export add-in may store a presenter's name or an output folder there.
- **Picture descriptions and links** that hold the full path of the file a picture came from.

## How to clean a presentation

1. Drop your .pptx decks on the box, or use **Choose files**.
2. Look at the top of each result. When it begins with the heading **Still in the cleaned copy**, the deck still holds the items named there; read them before you download.
3. Read the **Before and after** table: comments, comment authors, custom properties, template, preview image, photos with metadata inside, printer settings and speaker notes.
4. Note the message about speaker notes if your deck has them; they are kept. The **Removed** line shows what went, with a count for add-in tags.
5. Download the cleaned deck and open it once in PowerPoint, Keynote or Google Slides before sending.

## What is removed

- who created and last saved the deck, plus its title, subject, keywords, revision number and timestamps;
- the company, manager, template name and editing time PowerPoint recorded;
- the program that saved the deck and its version number;
- custom properties, and custom XML parts from templates or document systems;
- review comments, old-style and modern, together with the lists of people who wrote them;
- the first-slide thumbnail;
- the printer setup PowerPoint saved with the deck, including the printer's name;
- co-authoring change records and revision times;
- add-in tags, on the presentation and on shapes. The **Removed** line gives their number, for example "add-in tags (3)";
- sensitivity-label data, slide-library update records and the link list stored with the properties;
- the web address an online picture was inserted from, which PowerPoint keeps beside the picture (the picture itself stays in the deck);
- hidden data inside the JPEG, PNG, WebP and HEIC pictures on your slides: EXIF, GPS, XMP and IPTC. The pictures are cleaned at the byte level and look the same.

Folder paths are trimmed, not deleted. Where a link target, a chart's data reference, or a picture description made up of a path alone reads like `C:\Users\dana\Pictures\site-visit.jpg`, the cleaned deck keeps `site-visit.jpg` only. A linked file is then looked for beside the deck. A path that turns up in the middle of a longer description is not cut; the result lists it.

## What stays, and is named in the result

Some items are left where they are because the deck would lose content or stop working without them. If the deck has any of them, the notice **Still in the cleaned copy** says so and shows which part of the file holds each one:

- **A digital signature**, which carries the certificate of the person who signed, their name included, and the time of signing.
- **Add-in, macro and control data**: whatever an Office add-in saved in its own part of the file, plus custom toolbar buttons and the settings of controls on a slide.
- **Linked pictures, video, sound and objects**, and links that lead into a private network.
- **Addresses in e-mail links** on slides or in the notes. The address a link opens can differ from the words it is attached to.
- **Parts no slide or other part points to, and markup the tool does not recognize.** Since it cannot say what these hold, it lists them.

Handwriting and pen drawings (ink) are content. They are kept, a note in the result says so, and the notice lists the ink data, because it can record the time of each stroke and the pen or device.

When the deck's own text, on a slide or in the notes, contains a name its properties record (the author, whoever last saved the deck, a commenter or anyone else on record there), the notice names it as well, visible text and hidden text alike. It is your call, after checking, whether the name stays.

## Speaker notes are kept on purpose

Speaker notes are content you wrote, not data the program added. Deleting them automatically could remove material you need, so they stay, and the result shows the line: "Speaker notes are part of the presentation's content and were kept; delete them in PowerPoint if they should not be shared."

Notes are a common place for things the audience should not see: internal cost figures, talking points about a competitor, a reminder about a colleague. Before sending a deck outside your organization, open the Notes pane on each slide, or delete all notes with PowerPoint's own tool, described below.

## What else to check yourself

The tool does not change what is on the slides, so look at these in PowerPoint:

- **Hidden slides**, which are skipped in slideshow mode but sit in the file.
- **Off-slide content**: objects dragged outside the slide area, such as an old version of a chart.
- **Invisible objects** and items covered by other shapes. One case is listed for you: when an object hidden through the Selection Pane holds text, the result names that text as hidden text still in the file; a hidden chart counts with its titles and labels. The object itself is not removed.
- **Paths and intranet addresses written on slides.** They stay, like all slide text. The result notes them under **Good to know** when they contain a drive letter, a network share or a home folder, or a private host with `http://` in front; code comments and fragments of web addresses on a slide are not picked out.
- **Embedded video and audio clips.** These are not cleaned. A phone video inserted into a deck can carry its own location. Clean the clip on the [video metadata page](/remove-metadata-from-video/) before inserting it.
- **Footers and the slide master**, where names and dates are often typed once and repeated everywhere.
- **Embedded spreadsheets.** A chart made in PowerPoint keeps a small workbook with its numbers inside the deck, and that workbook can have an author name of its own. It is not opened; the result lists it with the other embedded files.

PowerPoint's own Document Inspector, opened from the File tab under Info and Check for Issues, can find and remove off-slide content, invisible on-slide content and presentation notes. Microsoft notes that it cannot remove pictures in the notes and does not detect off-slide objects with animation effects. After running it, PowerPoint keeps removing personal information automatically each time you save, until you change that setting.

Only .pptx files are accepted; save .ppt files as .pptx first. Password-protected decks and files over 300 MB cannot be read. A deck is also turned away, with a message saying it is too large, when its XML parts unpack to more than 256 MB together (pictures do not count toward that) when any single part or picture is over 64 MB, or when a part stacks its XML elements over 512 levels deep (the limit is 2,048 for a part handled in one pass). The same goes for a deck whose package holds more than 10,000 entries, and an element with more than 2,000 attributes (namespace declarations count). Should one stretch of text in a part pass about 4 million characters, the deck is cleaned and that part is named as not checked. Unpacked, the package as a whole may come to 512 MB at most, with pictures and every part the tool never opens counted in. If you share the deck as a PDF, export it from the cleaned .pptx and check the PDF's own properties with a tool that can [edit or clear PDF metadata](https://pd00.com/edit-pdf-metadata/). Slide photos in JPEG, PNG, WebP or HEIC lose their camera data. Videos, sound clips, TIFF or GIF pictures and embedded objects are not opened by this tool; the result says they were left as they are, and a video can be cleaned on the video page before you put it on a slide. Slide shows (.ppsx), templates and macro-enabled decks are turned away: save a copy as .pptx first.

## What this tool cannot do

- **Decode binary values.** A deck with a modify password stores a hash and a salt in encoded form. When they are well-formed and of the expected length they pass unread; anything else in those places is checked as text and reported.
- **Chase a default name through the deck.** A recorded author or company that reads "owner", "user", "guest", "admin", "administrator", "test", "author", "microsoft" or "unknown", whatever the capitals, is deleted from the properties and not searched for anywhere else, since themes and layouts use the same words. A name of just one or two Latin letters or digits, or one without any letter or digit, is not searched for either; the result says so when that happens. A name written in another script is searched for at any length. Comment authors' initials that short are skipped without any note, as the full name next to them is the thing searched for.
- **See through an unusual coding of a name.** A recorded name is searched for in the cleaned deck as written and with a closed list of escapes resolved: character references, percent-escapes and a `+` for a space in link addresses, JSON escapes such as `\u004A`, and differences of case, width or invisible characters. A name stored in any other coding, base64 for one, goes unnoticed unless it also stands somewhere in plain form.
- **Catch a folder path in an odd spelling within a longer text.** A whole value or a whole quoted argument that is a path is recognized in every spelling but two. On an old Mac disk with a name someone chose, only a path that reaches a file name with an extension counts (`Work Disk:Decks:Alice:talk.pptx` does, `Work Disk:Decks:Alice:talk` does not). After `File:` with a capital F, a slash or a drive letter must come next (`File:Users` does not count). Within a longer text, so are the everyday spellings: drive letter plus slash, network share, `file:` address, `/Users` and `/home` style folders, `~/`, environment variables, `Macintosh HD:`. Something like "C:x\Alice\notes" (drive letter, no slash) in the middle of a description may slip through, and so may a root that stands alone there ("~jane/", with no folder after it).
- **Judge what a picture shows.** Slide photos lose their hidden data, yet a face, a whiteboard or a screen in the photo is still there. The [redaction tool](/redact-image/) covers those.
- **Throw away the cropped-off edges of a slide picture.** PowerPoint keeps the whole picture when you crop it, and whoever gets the deck can drag the crop back. Cleaning leaves pixels alone, so the hidden edges survive. Before cleaning, select the picture, use **Compress Pictures** on the **Picture Format** tab with **Delete cropped areas of pictures** ticked, or replace the picture with a saved copy of the cropped version.
- **Beat a deck assembled to get past it.** A .pptx that someone builds by hand to carry data out unseen may contain a few forms the checks do not catch: text stored on a slide in the form another Office program uses; fields filled with content of a kind they never hold in a real presentation; markup nested or declared in ways the format forbids. Packages padded with thousands of repeats may also take noticeably longer to process.

When no item is listed, the tool found nothing it checks for; that is all the result claims. See the [limits](/methodology/#limits) for what it does not check.

## A short routine before a deck goes out

1. Run through the deck in Slide Sorter view. Hidden slides are marked there; delete the ones the audience should not have.
2. Open the Selection Pane, found under **Select** on the Home tab, on slides with complex layouts. It lists every object, including ones that are hidden or sit outside the visible area.
3. Read the notes for each slide, or remove all notes if the deck is going to people outside your team.
4. Check the slide master for names, dates and confidentiality labels that repeat on every slide.
5. Clean the final file here and send that copy.

## Where your presentation goes

The deck is unpacked and rewritten inside this browser tab. Slides, notes, comments and images are not uploaded, and nothing is stored after you close the page. An unreleased product deck can be cleaned here without leaving your computer.

## How the PowerPoint cleaning works

A .pptx file is a zip package. In a background worker, the page opens it with the fflate compression library (fetched from this site on your first Office clean), finds the comment and comment-author parts through the package's relationships, removes them and the relationships pointing to them, wipes what identifies people from the property parts (names, company, dates, the list of links) and deletes the thumbnail and custom XML. Every picture in the package is recognized by its contents, and JPEG, PNG, WebP and HEIC files are handed to the byte-level photo cleaner. Tag parts, co-authoring parts and the references to them are taken out, comments written into the XML itself are stripped, paths are trimmed, and what is drawn on the slides, the layouts and the notes are copied as they are. Everything is then written into a new .pptx whose zip entries carry a fixed date.

Before you see a result, the new file gets an inspection of its own, done by code that is kept apart from the cleaner. It opens the .pptx again and reads it part after part, testing relationships, elements and attributes against the published definitions of the presentation format. If a place meant for a name, an account id, an editing time, a path or add-in data is not empty, or if something is not in the definitions at all, it goes on the list you see. The result reports nothing left only when this inspection comes back empty, and even then it speaks for what the inspection covers, not for the whole deck. The [methodology page](/methodology/) covers how this is tested and how to confirm, using your browser's Network panel, that no file is uploaded.

## FAQ

### Does cleaning remove my speaker notes?

No. Notes are kept because they are part of the presentation. Delete them in PowerPoint if they should not be shared.

### Do photos on my slides really contain GPS data?

They can. A photo inserted from a phone may keep its EXIF block inside the deck. The **Photos with metadata inside** row shows how many images still carried metadata before cleaning.

### Will the slides look different after cleaning?

No. What is drawn on the slides is not changed, and images are cleaned without being re-encoded. Three things you cannot see on a slide do change: add-in tags are removed, a picture description that was a full file path is cut to the file name, and the name of the original file stored with a picture in an older-style drawing is taken out.

### Are comments removed or just hidden?

Removed. The comment parts and their author lists are deleted from the file.

### What does "Still in the cleaned copy" mean for a deck?

The tool kept something on purpose or could not check it: a digital signature, add-in data, a linked video, or a part it does not know. The notice lists each item and names the part that contains it, so you can decide whether to deal with it in PowerPoint before sending.

### Does this work for Keynote or Google Slides files?

Only for decks saved or exported as .pptx. Export to .pptx first, then clean the export.

### Can I clean several decks in one go?

Yes. Each deck gets its own table, and **Download all (zip)** saves them together.

## Sources

Checked 2026-10-11.

- [What the Document Inspector finds in PowerPoint (Microsoft Support)](https://support.microsoft.com/en-gb/office/remove-hidden-data-and-personal-information-by-inspecting-documents-presentations-or-workbooks-356b7b5d-77af-44fe-a07f-9aa4d085966f)
- [How photos store GPS: ExifTool's EXIF tag table](https://exiftool.org/TagNames/EXIF.html)

Last updated: 2026-10-17