Roblox is changing the way it handles bad language. The popular gaming platform just launched a new artificial intelligence feature that actively rephrases inappropriate words in real time.
In the past, the game used standard chat filters to block profanity. Whenever a player typed something against the rules, the system simply replaced the bad words with a string of hash marks (####). However, Roblox realized that seeing a screen full of symbols was disruptive and made it hard for players to understand the conversation.
The new AI tool takes a smarter approach. Instead of just deleting the offending word, the AI will rewrite the sentence using polite substitutes. Rajiv Bhatia, the Chief Safety Officer at Roblox, explained that the system is currently focusing on profanity. For example, if someone types “Hurry TF up,” the AI will automatically change the message to read simply, “Hurry up!”
To keep things transparent, the game will alert everyone in the chat when a message has been altered. The person who sent the message will also get a notification showing exactly which words the AI removed. Bhatia hopes this immediate feedback will actually teach users how to behave better and follow the community rules over time.
It is important to note that the AI edit doesn’t give players a free pass to swear. If a user constantly tries to send inappropriate messages, they will still face penalties or bans, even if the AI cleans up their text before anyone else sees it.
This feature is rolling out now, but it currently only applies to users who have completed the age verification process and are chatting with people their own age. Roblox has faced intense pressure recently regarding child safety.
After reports surfaced that adults were using the game to target kids, the company banned children under 13 from using general in-game chat. Despite these changes, the company is still fighting lawsuits from officials in Los Angeles and Louisiana, who claim the platform remains unsafe for minors.
