SI Agent News
DevelopingOpenAI

ChatGPT user says Enter still sends messages after the send shortcut was changed

A forum post describes ChatGPT sending a message on Enter even after the send key was changed. It is a small annoyance, but an accidental send can leak unfinished text.

The short version

Why it matters

Sending a half-written prompt by accident wastes time, and it can send private text you meant to cut. The report is narrow and unconfirmed at scale, but it is a good reason to check your own settings and build safer typing habits.

What the post describes

As community.openai.com reported, in a post on OpenAI's user forum, a ChatGPT user changed the message-sending shortcut in the settings to a key other than Enter. The expectation was simple: Enter would then add a new line, and the other key would send. Instead, pressing Enter in the input box still sent the message.

The poster says this was inconvenient because they often want a line break while typing. Hitting Enter mid-thought fired off an unfinished message. The poster describes themselves as a beginner on the forum and writes in Japanese with a translation tool, so some wording is rough, but the core complaint is clear.

The workaround, in plain terms

Rather than wait for a fix, the poster wrote a small script and ran it with Tampermonkey, a browser add-on that lets you run your own bits of code on websites you visit. The script watches for the Enter key. When it sees a plain Enter press inside a text-entry area, it stops the page from reacting and inserts a line break instead.

The script is selective. It leaves alone Enter presses combined with modifier keys such as Alt or Ctrl, and it skips cases where the user is in the middle of composing text with an input method, the kind of typing aid used for languages like Japanese. It also uses two different ways of inserting the line break, depending on what type of text box the page uses.

The poster reports that this worked: plain Enter no longer sent the message and added a new line instead. The post also points to a YouTube video, in Japanese with automatic English dubbing available, that shows the problem and the test. The code is in the video description, along with notes the poster asks people to read before using it.

The poster's own caveat

Partway through, the poster adds a twist. They write that it turned out the behavior could have been changed in the settings after all, and ask readers to treat the post as a record of messing around rather than a recommended fix.

That makes the story harder to read cleanly. The post does not say which setting solves it, or whether the original problem was a true malfunction, a misunderstanding of how the setting works, or something that only affects certain setups. It also does not say whether OpenAI has acknowledged anything. Our newsletter flagged it as an open user report, and that still fits.

Why a small bug still deserves a look

Keyboard behavior sounds trivial, but a chat box is where people paste drafts, names, numbers and half-formed questions. An unexpected send can fire off a prompt before you finished editing it, including text you wanted to trim first.

It also shows how far a frustrated user may go. Running custom code on a site you rely on can break things, and it can be risky if you copy scripts you do not understand. The poster did warn readers to read the notes before using the code, which is sensible advice for any script from a stranger.

What to do in the meantime

First, open your own ChatGPT settings and check how the send shortcut is set, and test it with a throwaway message before typing anything long or sensitive. The report may not apply to you at all.

Second, adopt habits that make accidents cheap. Draft long prompts in a notes app and paste them in when ready. Many chat tools use Shift+Enter for a line break, so it is worth trying, though this post does not confirm it for ChatGPT.

What to watch

Try it

Open ChatGPT's settings, check the send shortcut, then type a throwaway message and press Enter and Shift+Enter to see which one sends and which adds a line.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by community.openai.com · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news