ChatGPT Special Characters Removal — Clean Up Non-Standard Text
How to identify and safely remove special characters from ChatGPT text — smart quotes, dashes, symbols, and ligatures.
ChatGPT text often contains special characters that go beyond basic ASCII: smart quotes, various dash types, mathematical symbols, arrows, and typographic ligatures. While these characters look fine in modern browsers, they can cause problems in email, legacy systems, and any environment that expects plain ASCII text. Here is how to handle them.
What Special Characters Appear in ChatGPT Output
ChatGPT commonly produces: left and right double quotation marks (curly or smart quotes), left and right single quotation marks (including apostrophes), em dashes and en dashes, the horizontal ellipsis character (a single character instead of three periods), the multiplication sign (instead of the letter x), the minus sign (instead of a hyphen), various arrow characters, and typographic ligatures like fi and fl as single characters. Each of these has an ASCII equivalent that is more universally compatible.
Smart Quotes and Curly Apostrophes
ChatGPT frequently uses smart (curly) quotes instead of straight quotes. Left double quote is U+201C, right double quote is U+201D, left single quote is U+2018, and right single quote is U+2019 (this is also the curly apostrophe). These display correctly in most modern systems but can appear as garbled characters in systems expecting ASCII. Replace left and right double quotes with the standard quotation mark (U+0022). Replace left and right single quotes with the standard apostrophe (U+0027).
Dashes — En, Em, and Horizontal Bar
ChatGPT uses three dash types: the hyphen (U+002D, standard), the en dash (U+2013, slightly longer), and the em dash (U+2014, longest). The en dash is typically used for ranges (pages 10-20) and the em dash for parenthetical breaks. Both can cause display issues in systems that do not support them. Replace en dashes with hyphens. Replace em dashes with double hyphens or remove them depending on your preference. For details, see our em dash guide.
Mathematical Symbols and Arrows
ChatGPT occasionally uses the multiplication sign (x), the division sign, various arrow characters, and other mathematical symbols. These are Unicode characters that may not display in all environments. Replace the multiplication sign with the letter x, the division sign with a forward slash, and arrow characters with text alternatives like "to" or "leads to" or simple ASCII arrows like "->" depending on context.
Ligatures and Composed Characters
Typographic ligatures like fi (U+FB01) and fl (U+FB02) combine two letters into a single character. These can break text search (searching for "fi" will not match the ligature character) and cause font rendering issues. Unicode normalisation using NFKC form decomposes these ligatures into their component characters automatically. Apply NFKC normalisation to resolve all ligature issues in one step.
How to Remove All Special Characters Safely
The safest approach is targeted replacement rather than blanket removal. Replace each special character type with its ASCII equivalent: smart quotes to straight quotes, em dashes to hyphens, ellipsis to three periods, multiplication sign to x, and so on. This preserves the meaning of the text while ensuring universal compatibility. A text cleaner that handles ChatGPT-specific special characters does all of these replacements in one pass. For comprehensive cleaning, see our main cleaning guide and workflow guide.