Will there ever be an option for developers to opt out of this for their games?
Not only does it seem unnecessary, but this is not an appropriate way to deal with tagging/filtering in roblox.
There’s enough issues as is with roblox’s filtering system tagging out entire blocks of messages when the issue is only like, one word. How could I trust AI to not change the entirety of my message, when the filtering system is already difficult to deal with? I can only imagine this exasperating the issue of roblox’s chat system, especially for genres like roleplay games that depend on a functional chat.
I’m late, but I wanted to ask: are you guys stupid?
You’ve proven that AI is able monitor chat, even MODIFY user’s messages, then why do we have to go through all of this? Instead of improving the filter, yeah, restrict everyone from chatting until they show their face. You guys do not realize how many caveats there are with age verification, they overweight ALL positives.
From what I’ve heard “if behavior doesn’t match the age, user will need to re-verify”, what behavior do you mean? If you’re able to detect one’s behavior, can you not detect inappropriate behavior? This is beyond stupidity.
I never in 10 years of knowing Roblox hated it as much as I hate it now. I am staying away, off.
Don’t know if anyone else is experiencing this, but the filter has a massive problem with disability topics at the moment. It completely blocks saying “i have ADHD”, “i have autism”, and “i have autism spectrum disorder”. On occasion it has blocked saying “i have a disability” (included frick heck to show the rephrasing was being used actively). Rephrasing it on my own had it go through but this is a bit ridiculous. On the other hand, “serous cystadenocarcinoma of the ovary” goes through just fine for some reason.
Honestly with the age verification requirement you should just stop filtering for adults, or give an option to disable it.
(edited to correct grammar)
It has come to the point that the censorship is completely out of control. “Nap” is censored, but “naps” is not. Even the simplest, most basic words trigger replacement of all text with hashes, and since ther is no way to go up and edit the latest message, people simply add a “nvm” to it and go do something else. Not bad
Still not a fan of this police state thought crime kind of crap, nobody wants this and this should never become the norm anywhere. Give people autonomy over their own communication, let users review rephrased text before it is sent. Plenty of examples here already about how this AI model does not understand how tone and lingual nuance works in online communication. This will never reflect the user’s authentic voice and it’s wildly offensive to rob them of the ability to stop incorrect messages from being sent under their own name to their own friends.
Is this accessible via the API currently? If not, will it be?
Example; experiences that use custom chat interfaces (for example, radio-style UIs) while still complying with required services like TextChatService, will this feature be accessible in those contexts as well?
Why do you guys keep on making these kinds of updates? Why can’t you just listen to us? Many countries are planning to ban Roblox entirely, some have banned communication within the platform, and some HAVE banned their people from playing Roblox altogether! I live in the Middle East, and I can’t even test this update out since chat’s been banned; you aren’t making the ‘moderation’ better, you’re making it worse.
As a member of the developer community and a longtime player, this is thoroughly disgusting and disappointing. Others have made posts that outline my gripes with this–false positives, sanitization of social interactions, broken admin systems–but it can really be boiled down to laziness.
You will sooner drain oceans than invest in your moderation team.
That is embarrassing. Not just for you, but for all developers on this platform. Do better, or lose your userbase.
It doesn’t even work half the time… I’m not sure if it’s just the selection of games I play or what but the rephrasing thingy doesn’t do its job. A couple of times I saw the “(rephrased)” dropdown come up on messages that were strings of tags.
The message that would’ve been tagged if not for the rephraser…was tagged. You can’t make this up.
