ChatGPT is losing basic rich-text formatting when copying into Apple Notes and other apps. Bold text, headings, bullet points, and spacing disappear. Something that previously worked is now unreliable.
This is a serious regression in one of ChatGPT’s most fundamental features. Users shouldn’t have to rebuild their formatting every time they copy a response, especially when they’re paying for the service.
OpenAI needs to treat this as an urgent bug, not a minor inconvenience.
If you’re experiencing this too, upvote, comment with your affected apps, and report the bug to OpenAI. Let’s get enough visibility that this finally gets fixed.
How is basic copy-and-paste still breaking in a paid AI product?
This problem occurs with ChatGPT version 6, whereas working with version 5.6 causes no issues. I just tested it.
I am still encountering the same issues with 5.6 Sol. Everything was working perfectly fine until about 12 hrs ago.
I am still testing/working and experiencing this regression as well. At one point, version 6 was handling it perfectly, but now inline links are broken again.
The links are fully visible and clickable inside the browser UI (testing on both Brave and Firefox), but the moment you copy the response, the actual hyperlinks completely disappear from the clipboard data. You just get the plain text without the source URLs attached.
Actually, it’s not even just a problem with the links—rich text formatting as a whole is broken. Things like bold, italics, and headers aren’t copying over at all anymore.
Meanwhile, version 5.6 is running consistently and has absolutely no issues with this. Just wanted to give you an heads-up, hope it helps!
Thank You. I have the exact same issues, but on 5.6 Sol. I have created some work arounds for the time being, but we don’t have to jump through hoops to address issues as basic as this.
Copying a ChatGPT response using the built-in Copy button does not reliably preserve hyperlinks or text formatting.
Links are visible and clickable in the response, but the copied output contains only their display text - the destination URLs are missing. Formatting such as bold and italics is also lost.
This initially affected responses from version 6.0, but I’m now seeing the same problem with Sol 5.6, where copying previously worked correctly.
The copied content should preserve links and formatting, either as rich text or equivalent Markdown. Currently, what you see in the chat is not what you get when you paste.
Please investigate the native Copy functionality, particularly how it handles the newer response renderer and its custom link elements.
Update:
Sol 5.6 is working normally again in my latest tests, preserving links and formatting. The behavior has been inconsistent, so my earlier observation that 5.6 was also affected needs that clarification.
Version 6.0, however, consistently fails in my tests: links remain visible and clickable in the chat, but the built-in Copy button drops their URLs and loses formatting.
I am having the same problem is there any formatting or copying regression? Some old replies are showing no problem but the new ones are
Update from my latest tests with version 6: hyperlinks are now preserved when copying, but bold and italic formatting are still lost - including bold formatting applied to hyperlink text. The formatting is visible in the original response, but disappears in the copied output.
Have you compared the Copy button with manually selecting the same response, then pasted both into Apple Notes? A small example with a heading, bold text and a link could make the formatting differences easier to reproduce. That worked for me when I tried.
I tried the same and it didn’t work for me.
-Sriram Natarajan
Isn’t there any way to solve this problem? I am a paid user i don’t expect this from here i tried layout and visuals, web, ios and android no where it seems to be working. Even in Obsidian its failing to retain the bold formatting where it heavily relies on Markdown. Idk why in open ai no problem is showing in their status, I contacted them, doubt it would ever reach them due to high bottleneck