Roblox stopped hiding swearing behind ##### and started rewriting the sentence
Instead of swapping the banned word for a row of hashes, chat now rewrites the message into an accepted version, keeping the meaning. The punishment for swearing still exists. Here is what changed, what did not, and why everyone in the conversation is told.
Produced with AI from the sources cited below, and checked by automated verification against those same sources. No person read this text before publication. Editorial Policy
Anyone who plays Roblox knows the row of hashes. You type something in the heat of a match, hit send, and what appears on screen is #####. Your team does not understand the warning, the play is lost, and nobody learns anything from it.
The platform has changed that behaviour. Instead of hiding the word, chat now rewrites the whole sentence into a version that passes the rules. This piece explains what actually changed, what stayed the same — and the part almost nobody notices: everyone in the conversation is told there was a rewrite.
The short answer
Filtered text is now translated into more respectful language instead of turning into hashes. At launch, the rewriting covers profanity.
The example the platform itself gives is direct: someone trying to warn a friend about a monster behind them would type a sentence with swearing in it and the friend would see only a row of #. With rewriting, the sentence arrives with the meaning preserved and without the swearing.
Two things that did not change, and they are the most important part:
A rewritten message is still a violation. The same rules apply, and anyone who keeps swearing faces the same consequences as before. Rewriting does not forgive; it only stops the conversation from dying.
Everyone is told. When a message is rewritten, everyone in the conversation gets the notice that it was rewritten. It is not a silent correction that makes you look polite — it is a public one.
The platform describes its own system as a steering wheel rather than a stop sign. That is the difference: one tells you to stop, the other corrects the direction and lets you carry on.
How this shows up on your screen
There is nothing to switch on. The change applies to chat inside experiences and arrives on its own.
What you will notice, in order:
- You type the sentence normally and hit send.
- The sentence appears different from what you wrote, with the word swapped for another that passes the rules.
- A notice appears for the conversation, saying that text was rewritten.
If you write without swearing, none of this happens and chat works as it always did.
It is worth remembering there is a layer before this one, live since May 2025: the real-time warnings, which appear when you break a rule and ask you to reconsider your language next time. They apply to both text and voice, and they still exist.
Before and after, layer by layer
The table separates what changed from what stayed the same, and carries the two numbers the platform published from its experiments with warnings and time-outs: a 5 per cent drop in filtered messages and a 6 per cent drop in consequences coming from abuse reports.
They are small numbers, and it is worth reading them that way. A 5 per cent drop in filtered messages is not the problem solved — it is a sign that the warning changes the behaviour of some share of people. Rewriting is the next bet, and it has no published number yet.
It is worth understanding the order the layers act in, because they do not replace each other: first the filter decides the message breaks a rule; then the rewrite swaps the word for an accepted equivalent; in parallel, the real-time warning explains to the writer what was broken; and if the person persists, the consequences come. None of those stages was removed — rewriting slotted in between the second and the third.
The platform also says why it insists that the rewritten sentence appear for everyone and not just the writer: that way everyone in the conversation sees which language is accepted there. The writer sees what is not accepted; everyone else sees what is. It is a way of teaching the rule without interrupting the match — and it is very different from a discreet correction only the author notices.
The catches
Rewriting is not permission. This is the most common confusion. If you read the change as "now I can swear because the system fixes it", you will be punished all the same. The violation is still recorded.
The sentence that comes out is not the one you wrote. For anyone who uses chat creatively, that grates: your message can appear with a word you did not choose. There is no undoing it and no asking for the original version.
Everyone sees there was a rewrite. If the idea was to disguise it, it does not work. The notice goes to the whole conversation.
At launch, the scope is profanity. Not all filtered text is profanity, and what falls under other rules is still handled as before. If your message vanished entirely instead of being rewritten, it probably fell under a different rule.
Filters get it wrong, and still will. None of these layers guarantees an innocent word will not be blocked. The change improves what happens after the filter catches something, not the filter's aim.
What to do next
In practice what changes for you is small and good: you can warn your team about something urgent without the message turning into hashes. Use that.
If you tend to pick up filter warnings, it is worth understanding that your sentence will now keep appearing — rewritten and flagged. For anyone worried about their own account, the route has not changed: less swearing, fewer recorded violations.
And if you play with younger people, or manage a younger person's account, this change talks to the rules about who can speak to whom. Roblox chat has been through several changes in a row over recent months, and rewriting is just the most recent layer of a system that already had a filter, a warning and a punishment.
For anyone managing a child's account, there is a useful side effect here: before, a conversation full of hashes told you nothing about what was being said. Now the rewritten sentence appears, and with it you can get a sense of the tone of the conversation the child is in. That does not replace parental controls, which are still the place to limit who they talk to — but it is information that simply did not exist before.
And if you want to understand the subject to the end, the next question is what happens to someone who persists. The short answer: the consequences have not changed, and rewriting is not a second chance. It is only a way for the conversation to carry on while the rule is applied exactly as before.
Sources
Ready-to-use artifact
| Layer | Before | Now |
|---|---|---|
| Banned word | turned into ##### on screen | is rewritten into an accepted version |
| Who sees the change | only the writer understood | everyone in the conversation is told |
| Meaning of the sentence | was lost | the announcement says it stays closer to the intent |
| Real-time warning | has existed since May 2025 | continues, in text and in voice |
| Punishment for persisting | existed | unchanged |
| Scope at launch | — | profanity |
| Test measurement | Result |
|---|---|
| Filtered messages | 5% drop |
| Consequences from abuse reports | 6% drop |
Figures as stated in the official announcement of 05/03/2026, checked on 23/09/2026. Source: about.roblox.com/newsroom/2026/03/rethinking-chat-for-fun-gameplay-civility
