How do I stop Confluence adding HTML tags to image subtitles?

[Caolan Radford]

Every time I edit a page, Confluence adds a bunch of HTML tags to my image subtitles:

I added none of these. Every time I edit the page, more appear and the subtitles become increasingly unreadable. 

Things I have tried:

• Different browsers/updating my browsers/updating my pc
• Writing a new page (Happens on every page I edit)
• Restarting my PC

[Vishalakshi Narayana Swamy · Accepted solution]

Hi @Caolan Radford

Welcome to Atlassian community!

I see that when editing a confluence page, the image captions are rendering HTML tags in your instance.

I tried to replicate the issue from my end and I was unable to replicate the issue since the added images and their captions are working fine at my end.

In this we might need to investigate further for verifying the page content and the attachments by replicating the issue on your instance.

Since you are not an admin, can you advise one of your admins to raise a support request here https://support.atlassian.com/ so that we can assist you further.

Best Regards,

Vishalakshi Narayanaswamy

[Susan Ngo]

I suppose the issue appears to be resolved now? Although it appears this is at the cost of needing to go back and manually edit pages again that previously had this issue to remove all the existing HTML tags and any escaped markup, if any, and resave.

Saving now, does not appear to automatically add any of the initial paragraph tags anymore at least.

Thank you for this follow up post!

[Caolan Radford]

It’s all good. If it was worth their time, I’m sure they would’ve logged it already. Maybe let regular users report bugs instead?

Anyway, looks like the issue’s fixed — I can remove tags now and they stay gone. Appreciate that!

Also, kinda weird that you’re comfortable posting random admin emails in a public forum?? Can you remove them from your comment? Feels like you’re playing it a little fast and loose with private data there, no?

I don’t want to endorse an answer that exposes private information. Thanks.

[Caolan Radford]

Appreciate the edit!

[Sarah Schallhorn]

Backing-up @Caolan Radford on this.. no way this is working as-intended @Mia Tamm

Just created this page and added plaintext captions. After two edits it's a total mess

[Susan Ngo]

Submitted feedback regarding this as well since was not sure how to go about reporting this bug/issue. A colleague brought it up to my attention today and I tested in 2 separate workspaces. This appears to be an issue starting today as I did not encounter this issue yesterday when I was editing a page with an image that had a caption. The moment I edited the same page I was working on yesterday, it added the html paragraph tags (<p></p>) to all images on the page (existing and new ones). Making additional edits without clearing all the paragraph tags would add more escaped markup.

The following is basically what it ends up appearing as after saving despite when editing, the caption only read as "Interactive bar graph":

No amount of editing to try to manually remove the html tags will actually remove it since Confluence appears to automatically and continuously try to add additional paragraph tags to the caption.

[Shawn Doyle - ReleaseTEAM]

Hi @Caolan Radford 

I think you have found a bug.

I see the same thing in my Confluence instance.

[Caolan Radford]

!! Dang.. Seeing as I am not an admin on our system, I don't have rights to open a support ticket?? This completely breaks old documents, making any editing impossible. The product is pretty unusable until this gets sorted

[Shawn Doyle - ReleaseTEAM]

I've asked support to become involved.  Hopefully they will address this quickly.

[Mia Tamm]

Hi @Caolan Radford

Confluence is actually behaving as designed: the caption field is plain-text, so any <p>…</p> you type gets escaped to &lt;p&gt;…&lt;/p&gt; and then wrapped in its own paragraph on save. There’s no “disable HTML escaping” toggle in Cloud or the new editor, so to keep your subtitles clean, here’s what I recommend:

1. Use plain text (or wiki markup) only.
   • Edit the image, strip out all raw HTML tags, and leave only the human-readable text, for example:
     

     Use this to flip between modes in the inspector

   • Confluence will still wrap that in a single <p>…</p>, but you won’t see the double-escaped tags.
     
     
2. For rich formatting, use built-in markup.
   • To bold or italicize, use *bold* or _italic_.
   • For links: [Link Text|https://example.com].
   • This way Confluence applies the correct HTML behind the scenes without ever showing you the tags.
3. Server/DC only: raw-HTML workarounds.
   • HTML Macro: Ask your admin to enable it, then wrap your caption in an {html} macro so Confluence won’t escape it.
   • Layout hack: Insert a 1×2 table or panel, put your image (no caption) in one cell and your custom-HTML caption in the other—again inside an HTML macro.

Unfortunately, in Cloud and in the new editor there’s no hidden switch to turn off escaping in captions. The simplest way is to lean on Confluence’s wiki markup or keep your caption as plain text—and let Confluence add that single clean <p> for you. Hope that helps!

— Mia Tamm from Simpleasyty

[Caolan Radford]

No you misunderstand. I did not add any tags. Confluence adds more every time I hit 'edit'. This happens on pages from years ago, if I hit edit, new tags appear and fail to get rendered properly, completely destroying readability on any page that has plain text captions that wants to get edited.

[Mia Tamm]

Hi @Caolan Radford — sorry for the frustration. Here’s the TL;DR:

1. What’s happening
   Confluence Cloud’s editor treats the image caption as plain text. Each time you save, it wraps whatever you have in a <p>…</p> and re-escapes any existing HTML entities—so old <p> tags become &lt;p&gt; and then get wrapped again, snowballing over repeated edits.
   
   
2. Why there’s no “off” switch
   This is by design in the Cloud editor—there isn’t a hidden toggle to disable HTML-escaping in captions. The caption field only accepts plaintext or wiki-style markup, which it then converts to its internal storage format.
   
   
3. Quick workaround
   
   
   1. Remove the caption: Edit the image, clear out the caption field completely.
      
      
   2. Re-add as page text: Directly under the image, type your caption as normal wiki text (or in a tiny 1×2 table/panel if you need special layout). That text won’t be re-escaped on save.

Hope that makes things clearer—and keeps your captions clean going forward!

[Caolan Radford]

I think its pretty clear this is a bug. Will await support. Not going to go editing years old captions to workaround this. Appreciate the effort tho!

[Mia Tamm]

Hi @Caolan Radford —thanks for the update! You may well be right that this is a bug.

If you’d like, please reach out to Atlassian support so they can investigate and address it.

Thanks again for flagging this!

[Caolan Radford]

I do not have rights to create a support ticket because I am not an admin, this post is the best I can do as far as I know