Roblox launches AI chat filter to rephrase offensive language
2026-03-06 - 07:43
Roblox is launching a real-time, AI-powered chat rephrasing feature that automatically replaces offensive words with more appropriate language, the company announced on Thursday. This new system enhances Roblox’s existing text filter, which only replaces banned words with “#” symbols. However, Roblox points out that strings of “####” can disrupt conversations and make them hard to follow. Now, instead of just showing hash marks, filtered texts will be rephrased into respectful language that closely matches the user’s original message. For instance, “Hurry TF up!” — previously shown as “####” — will now be rephrased to “Hurry up!” and everyone in the chat will be notified of the rephrasing to maintain civility. “Chat is central to how people connect, coordinate, and play on Roblox,” said Rajiv Bhatia, Roblox’s vice president of User and Discovery Product. “Real-time rephrasing helps keep gameplay and conversations on track while guiding language toward what’s appropriate, reducing friction without compromising our community standards,” Roblox emphasizes that while rephrasing lessens chat disruptions, its safety system still addresses more serious misconduct. The feature supports all languages currently available through Roblox’s auto-translation tools. The company is also upgrading the text-filtering system to detect more variations of banned language. Roblox says early results show that the system is getting better at detecting leetspeak and more sophisticated attempts to bypass filters. This has reduced the prevalence of false negatives when sharing or soliciting personal information by 20x. Thursday’s announcement arrives shortly after Roblox introduced compulsory facial verification for access to chats on its platform following a wave of lawsuits over child safety from the attorneys general of Texas, Kentucky, and Louisiana, among others. Lawsuits were filed against Roblox following reports that the platform was allegedly exposing young users to significant dangers, including explicit content and grooming.