Why does ChatGPT paste into Word as asterisks and hash symbols?

ChatGPT writes in markdown, where double asterisks mean bold and a leading hash marks a heading. Word does not read markdown, so it prints those symbols literally. Converting the markdown to rich text first makes Word receive real bold and heading styles instead.

How do I keep formatting when pasting ChatGPT into Word?

Put rich text on the clipboard instead of raw markdown. Paste your answer into a converter that renders the markdown, then copy and paste into Word. Choose Keep Source Formatting from the paste-options button to preserve the styles exactly.

Is pasting rich text better than exporting a .docx from ChatGPT?

For most edits, yes. A .docx export is a separate file you have to open, restyle, and merge. Pasting rich text drops the content straight into your open document at the cursor, keeping your existing template and styles.

Does this also fix the weird spacing from ChatGPT?

Yes. AI output often carries non-breaking spaces and zero-width characters that leave uneven gaps in Word. The converter strips them out automatically, so the pasted text spaces normally.

Does it work in Word for Mac and Word Online?

Yes. The copy is standard HTML rich text, which every modern version of Word reads: Microsoft 365, Word 2016 and later, Word for Mac, and Word on the web. A normal paste keeps the formatting.

← Back to Blog

How to paste ChatGPT into Word with formatting (2026)

You asked ChatGPT for a report, and it came back beautifully structured — bold section titles, numbered steps, clean bullet points. Then you pasted it into Word and got a wall of **asterisks** and # symbols where the formatting used to be. This is the single most common complaint about moving AI text into a document, and the fix takes about ten seconds once you understand what is actually happening.

The short version: paste your draft into the free ChatGPT to Word converter first, then copy from there into Word. It puts rich text on your clipboard instead of raw markdown, so the bold, headings, and lists arrive as real Word formatting. Here is why that step is necessary.

Why ChatGPT pastes as literal `**` and `#`

ChatGPT does not write formatted text. It writes markdown — a plain-text shorthand where **double asterisks** mean bold, a leading # marks a heading, a - starts a bullet, and 1. starts a numbered list. The ChatGPT interface quietly renders that markdown into visual bold and headings on screen, so you never see the raw symbols. Your eyes are looking at the rendered output, not the underlying text.

The moment you select and copy, though, you grab the underlying markdown — not the rendering. Word is a rich-text editor, not a markdown parser, so it has no idea what those symbols mean and prints every one of them literally. That is why a tidy ChatGPT answer lands in your document as ## Executive summary and **key finding** with the marks fully visible, and why every bullet shows a leading hyphen.

Google Docs, Gmail, and Outlook behave the same way for the same reason. Docs occasionally auto-detects a lone ** pair and converts it, which is arguably worse — you end up with inconsistent, half-formatted text that is even more work to clean up.

The hidden characters make it worse

On top of the visible symbols, AI output usually smuggles in invisible characters — non-breaking spaces and zero-width joiners — that survive the paste. They leave uneven gaps between words, so even after you manually delete every asterisk, the spacing still looks slightly off. If you have ever wondered why pasted AI text never quite sits right, our guide on why ChatGPT text has weird spacing breaks down the exact characters responsible.

The fix: put rich text on the clipboard, not markdown

The reliable solution is to convert the markdown into real rich text before it reaches Word. When the thing on your clipboard is HTML — the same format Word uses internally — Word reads it natively. Headings become Heading styles, **bold** becomes genuinely bold, and lists become real bulleted or numbered lists. There is nothing left to clean up.

That is exactly what the ChatGPT to Word converter does:

Paste your ChatGPT answer into the editor. Markdown symbols and hidden characters are stripped as it renders.
Check the preview — headings, bold phrases, and lists should look right. Fix anything with the toolbar.
Click Copy, switch to Word, and press Ctrl+V (Cmd+V on Mac).
If the paste-options button appears, choose Keep Source Formatting to preserve every style exactly, or Merge Formatting to adopt your document's own styles.

Because the copy is standard HTML, this works in Microsoft 365, Word 2016 and later, Word for Mac, and Word on the web.

Why not just export a .docx from ChatGPT?

ChatGPT can hand you a .docx file, and for a whole standalone document that is fine. But for the everyday case — dropping one answer into a report you are already writing — it is slower. The export is a separate file you have to open, restyle to match your template, and then copy across anyway. Pasting rich text skips all of that: the content lands exactly where your cursor sits, inside your existing margins, theme fonts, and section headers, and you keep typing.

One honest limit: complex tables and embedded images do not ride along on the clipboard. Add those with Word's own tools after the text is in. Everything else — bold, italics, underline, headings, quotes, inline code, and both list types — transfers cleanly.

The same fix works in Docs and email

This is not a Word-only problem, and it is not a Word-only fix. Anywhere ChatGPT markdown shows up as raw symbols, the same clean-copy approach solves it:

Cloud documents: ChatGPT to Google Docs handles Docs' inconsistent markdown detection so you get uniform formatting.
Email: the Gmail text formatter drops AI drafts into the compose window as clean rich text, with no stray asterisks in the message your recipient sees.
Anything else: the general-purpose AI text formatter adapts to whatever destination you paste into.

The pattern to remember is simple. ChatGPT gives you markdown; most apps want rich text; the converter sits between them. Run every AI draft through it once and you never delete another asterisk by hand.