Roblox VC Ban discovery chart

Overview

Forum Staff, if this is not the place for this topic, let me know so I can move it accordingly.

The current Voice moderation system is broken, providing no details to why you got banned/suspended, leaving users guessing what the reason may be. Within the last ~3 months, Voice V1 Settings endpoint now provides a numerical value dubbed “banReason”. No additional information other than this numerical variable is provided. The goal with this post is to gather information with suspected reasons to piece together what each banReason may be, allowing us to be more mindful to what was said/done and comply with the ToS.

How to contribute:

  • When suspended, be mindful of what you could have said to trigger the suspension.
  • While logged in, click on this link to view your voice settings.
  • Take a screenshot of the text/JSON data and reply to this thread with the following:
    • Endpoint Data:
    • Suspected reason: Why you think you got moderated. Specific details as to why (while ensuring you’re not violating Devforums ToS) is highly encouraged so we can attempt to identify triggers to avoid making repeat mistakes.

What is the purpose of this?

So we can attempt to figure out specific words, phrases, and contexts as to what is leading to VC moderation. This research is NOT INTENDED for evading Roblox moderation, but be a point of reference as how we can be compliant with Roblox rules and not blindly continue to violate the ToS.

The banReason variable

I am not sure when this was added to the endpoint, but with it being a recent addition, it gives me hope Roblox is making appropriate changes to provide us some details as to why the moderation occurred. For the meantime, we need some sort of solution allowing collaboration with Roblox while they figure this out. I have no doubts this is a priority for Roblox with VC being a major new feature, but with any new feature allowing opinion comes difficulties in isolating intentional bad actors who want to abuse whatever they can.

Ban chart:

BanReason 0:

  • You are not banned.

BanReason 5:

  • 2 account recorded.
  • Likely harassing/bullying others.

BanReason 6:

  • 2 account recorded.
  • Likely self-harm/threatening others.
25 Likes

To lead by example, I’ll start with mine:

  • Time occurred: Fri Feb 09 2024 20:10:52 GMT-0600 (Central Standard Time)
  • Endpoint Data:
    image
  • Suspension time: 18 minutes.
  • Suspected reasoning: Unknown. I was making small talk at the time, and the suspension occurred when I wasn’t even talking.
6 Likes

Replies are disabled after about 10 minutes in #bulletin-board. The intent is for members to post their ban time and reason they suspect they got banned in order to try and correlate what each numerical banReason is. If I was to mediate this information through private messages, the overall quality would be subpar compared to open collaboration.

5 Likes

Filed on behalf of Larcenyl

  • Time occurred: Sun Feb 11 2024 22:52:?? GMT-0500 (Eastern Standard Time)
  • Endpoint Data:
  • Suspension time: 5 minutes.
  • Suspected reasoning: Saying without direction for someone to game end (using alternative language for the sake of the thread).
2 Likes

Let me start by saying, this might be a very long read, and I don’t know if it’s against Global Rule 8; As it’s talking about Roblox moderation outside of DevForums so I assume it’s okay.

I’ve always wondered about Roblox voice moderation. I’ve worked inside of content moderation for some corporations, Facebook / Meta, and Gaggle (Children Safety K - 12) for about a year in each company; so I feel like I know a little about it.

Let’s start by summing up what Roblox voice moderation is: It’s an AI called “Speechly” that joined Roblox around September 2023 (They bought them out). It’s a "fast real-time auto-moderation system , like how COD uses “Modulate” or “ToxMod” for their AI moderation, granted ToxMod also watches chat 24/7 unlike Speechly, which is for VC only (This plays a key part to some odd bypasses you can do to Speechly).

Voice content SHOULD adhere to Terms of Use (TOS) and Community Standards, and the 17+ Policy Standards make everything even more confusing; The main one you should be looking at is the “Community Standards” as the 17+ is more directed to experiences and not said content besides swearing (They at least updated it December 5, 2023). This is how I read it, and I see no other way to read it when the “Chat With Voice” page only mentions 13+ (Feel free to correct me as Roblox support couldn’t answer it when I got my ban; They said “We’re sorry but we’re unable to provide you with any further information or response regarding this inquiry.”)

The great thing you’ll notice is Speechly and most other auto-moderation tools are almost like “Nineteen Eighty-Four by George Orwell” or a “Big Brother System” however you know it; They work well, like really well, like it or not; especially when a platform contains kids and adults.

The big problem that you’ll find with a system like this is the occasional odd bans / suspensions that don’t really make sense, I bet you’ve seen it even on Facebook / Meta’s auto-moderation; I can give you a pretty good reason why - It’s following a TOS / Community Standard to a T and doesn’t care about the context of said thing. It’s even happened to me on really weird things that are not mean or directed at a single person, just that what I said would semantically be against TOS.

I feel like I can’t really elaborate on this as I will be disclosing how to bypass a system when there are children on the platform; Which leaves a bad taste in my mouth, but I can give some fun and weird examples that have directly happened to me - I can evolve on one, just to have some attention brought to it, as personally I think it’s kind of funny, but also BAD; Saying “I love ___”, but you can say “ily” inside of text chat, and even “I love ‘username’” but when you say it in VC you can get banned; This also includes the words: Date, Flirt, Love, Marry, and many others that can be deemed under the “Romantic and Sexual Content” Community Guidelines Roblox has laid out, as previously said, the 17+ Policy Standards are due more to them being towards “Experiences” and not players / chat. At the end of the day, I’ll say it; and I’ll say it again, keeping “Romantic and Sexual Content” off the game, is the most important thing when we mix adults and minors onto the same game platform, even if the bans feel unfair; When you’re an adult you have to keep in mind what you’re saying towards others in real life, the same should be said about Roblox as a standard, 17+ is just a mitigation to keep younger people off, not a cure; Always keep in mind what you’re saying unless it’s cursing in a 17+ game, it’s cool I guess.

I’ve talked to 45 people within the last few days inside a 17+ server (Neighbors and Mic Up) and 5/6 say the same thing, banned for something that could be deemed flirtatious even if it’s about themselveS. The people I talked to were also banned previous to the VC suspensions / or got a longer suspension than the 5 minutes on; The people I talked to also seemed to get more frequent bans if they have one previously. Overall, it seems that Harassment / Stalking was the top ban in VC / Account suspension, most people defended themselves and said there was more than one report; When I got my ban it was probably 35+ within 3 minutes, so it could lead to mass reports = bans.

The other 1/6 I talked to about bans / suspension couldn’t seem to remember it, or said it was 2023 prior to ban; which would put it within prior December update of rules? (Please someone give me better dates, I can’t find it.) Overall, the Roblox AI moderation will continue to be a faulty mess, and we can’t do anything about it. The best thing we can do is watch our language and adhere directly to the TOS, even as stupid as that sounds, just keep every conversation to a T rating to be safe.

  • Hopefully this will give you a insight inside of the AI and not Roblox, I’m always curious of responses, and will answer anything that you may think is wrong.

  • This is my opinion and nothing farther as someone who talked to people ingame and my own experiences ingame.

  • There is even more weird bypasses you can do, but I will not say them here, I am not advocating for an AI monitored text chat, but I want better separation between minors and people older(17+);Until this - the system they have in place is to stop people from “flirting” and malicious content with minors and this includes the (Weird)UGC bans if you read the TOS; Even the random ones.

  • What to take in: Roblox has a system in place to protect minors even inside of a 17+ games beyond swearing; next time you use VC you just have to be very careful on voice that could endangers minors / semantically void the TOS.

TLDR:I’ll say is that at the end of the day, Roblox is trying to make a community that is supportive of each other; When you’re in a 17+ Game you just need to be mindful of being Flirtatious or Malicious in any matter, as it’s semantically against Terms of Use

- One last thing, just because you see something against TOS or Community rules; Doesn’t mean you should call it out, that is probably the worst thing to do, at least in via voice chat, always use the report function. :smile:

Edit: Making everything a smoother read

4 Likes
  • Time Occurred: 2/18/2024 @ 2015 CST
  • Endpoint Data:
    image
  • Suspension Time: 20 minutes
  • Suspected Reason: Telling someone no one cares in a sarcastic manner.
2 Likes

Hello, I got a message from the forum moderation team saying to flag this post as they thought I was ‘minimodding’ when I didn’t flag this.

So yeah, should I flag this? This is useful, so I don’t see any reason to do so.
Anyways, the post I made was no longer relevant and it’s more efficient to just submit a feature request to give context on VC ban reasoning, since it’s likely to be reviewed by staff.

This endpoint has been known for ages, I saw another post on it.

5 is related to threats that involve violence or other topic roblox deems inappropriate.
6 is related to Profanity / Harassment.

2 Likes

The endpoint is known, yes, but the BanReason has been a relatively recent addition. Can you provide the context for confirming these two BanReasons?