[Please, staff attention and response only] How can I protect my account from unintentional possible use of banned words?

Good day!

I have a bot hanging on my main account that interacts with people and can produce arbitrary responses triggered by outside people. Here are some examples:

  1. The bot uses, trivially, an arbitrary command system where people can score their own command names and responses to them.
  2. The bot can add tracks to a Spotify queue via channel scores and control the player via chat commands.

So here goes. In the first case, I’ve seen a bunch of suicidal people who don’t care about their account (or, they can register a new one) and they break the rules left and right and can add a command with forbidden words in response, and my bot will then write them in the chat. In the second case, Spotify has tracks with banned words on Twitch in the title, and my bot can write the track title when requested via the appropriate command or when ordered via channel points.

I work, in priority, for Russian chats and I can, in theory, write a word filter in Russian. If I cover every language on the planet, I’ll just go crazy, you understand. Surely there must be a gentler way.

So, my question is: how can I tell Twitch that I may not have intentionally written a banned word, but that it happened due to a request or actions of a third-party user, so that my account doesn’t get banned on the platform?

First, you are unlikely to get a staff reponse, since it’s christmas break and even then what would be written could be mistaken as policy and if you do get banned you’d just finger point over here with “well x said it was ok” and it turned out x was wrong. Especailly since the staff that are here are not part of the policy team(s) related to your query. Since your query isn’t a developer query but a moderation query.

So the key points here are:

a) there are not “banned words” set by Twitch itself every channel is different and words not allowed in one channel by the channel moderators would be perfectly allowed in another. a lot of channels put fuck on their banned work list and a lot of other channels allow fuck for example
2) Context is king, it’s kinda obvious that if a given word is part of a song title, it’s part of a song title.

So the channel will run it’s own auto mod filter which no one can get around (mods included) so the channel can set the banned words in the list which would stop the bot posting.

or you provide a bot panel that includes a way to channel moderators to provide banned words and thus stop the bot posting

To my knowledge Twitch doesn’t have a list of “chat banned words”.

Sure Twitch blocks certain words from use in polls/predictions/titles/etc but that doesn’t apply to chat and shouldn’t result in the a user account using those words in chat from being banned from the service

You can’t as any given account can be used as a regular account, credentials could be stolen/etc there isn’t a special “bot account type”.

Basically bots and people using third party chat apps, like chatty for example, all login to chat the same way and cannot be distinighiused as such by Twitch.

No on knows if a given message sent by an account is from a human, a bot, or someone using a third party chat app, or some random script, or the account details are stolen.

Normally I’d point you at Twitch Support here but they’d just send you back here. Basically the problem you are trying to solve doesn’t really exist and you cannot be elevated above the law.

If your chat bot is clearly posting a song title as a song title (clearly tagged as a song/now playing maybe also with the spotify link/etc), and the user account gets reported for it, then you shouldn’t get banned as it should be obvious to the report reviewer the context of the chat message.

I have seen people claiming their bots were banned for commands that said things like “I’m 12”. That’s not a channel moderation thing but a Twitch thing.

At the end of the day, you’re responsible for the account. If it posts content that violates Twitch’s ToS/Community Guidelines then action may be taken.

As a developer, you are responsible for taking reasonable measures to prevent the bot sending messages that may violate the Community Guidelines. As Barry has said, context matters, but while that may help your case if a streamer uses special characters to disguise the use of making the bot send content that violates the guidelines, it may hurt your case if the bot sends a plain text slur that could have easily been prevented if the developer took reasonable precautions.

Act reasonably and responsibly, and in most cases you should be fine. But if the bot is a free for all that can send anything then it’s likely only a matter of time before it’s banned.

Good afternoon, by “staff” I meant you and other moderators first of all, because you know the answer to my question better than other developers. Thank you very much for answering and sorry for such a long response on my part.

So, look, my understanding is that there are still banned terms on Twitch where a single report would be enough and a person would be banned in a couple seconds. I’ve personally sent a report on violators and video one letter in the mail with the receipt of a complaint and a couple of seconds later another where the person is already banned. For example, in the CIS region, it is definitely 3 very popular words that: insult gays, dark-skinned people and people with disabilities (I probably will not write them, so as not to be banned because of the topic on the forum, where I wonder how to avoid this😆). These 3 words are already some kind of setting in people’s minds that they can’t be said unequivocally. But, for example, that Twitch can ban in situations where a person writes that he is, for example, 8 years old, which does not yet give the right to register and use the site - no, and such people fly to a permanent ban.

Next, you made me very happy when you said that content matters. In case of spotify-bot, indeed, it writes that it added a track to the queue, writes artists, track name, how long it lasts and a link to the track in spotify. So, if I understand correctly, I should not be banned if the bot, for example, writes that he put this track in the queue: https://i.imgur.com/o6bA9d8.png ? The title of this track is the 2 popular banned words in CIS mentioned above, for which they are instantly banned. If I write them in context, where it is perfectly clear that I’m not just insulting someone, and that I just put the track in the queue, but this track has just such a title?

You also say that my problem, allegedly, does not really exist, but if you take a system with custom commands, where third-party people can put their answers to the command, where already by the context, it will be considered as if I am offending someone, although my bot just responded to the command, which was hammered into it with forbidden words. What should I do here? I can still write a filter in Russian, but, suddenly, it will make an answer with a forbidden word, for example, in Spanish. I can’t download all the words in the world and use them to track banned words. I would just go crazy and it would take up a huge amount of disk space. What should I do if I just want to help some streamers and make a system of custom commands, but so that I don’t get banned if a third party writes a banned term in response to a command. So there’s no way?

Good afternoon, thank you so much for replying and sorry for such a long reply on my part.

Yes, I understand. But look. And if you take “world” famous bots like StreamElements, Nightbot, Moobot, Fossabot, etc. No, I understand, where I am before them😆 but how do they do in such situations? Do they have filters for every language on the planet? The management of Twitch must be sure that if these bots write a forbidden term of a particular region in the chat, it means that such words were set in the team either by the streamer himself or by his moderators, but not by the developer himself on behalf of the bot, so that appropriate action is taken against this bot.

And what should I do if I have a narrowly focused bot for several channels, but not for the whole Twitch? Should I download thousands of dictionaries and write thousands of filters in every language on the planet, and take into account which terms are forbidden in a particular region? Is this the only way I can protect the bot?

I also remembered one other thing. Twitch has a form to verify your account as a bot. According to the documentation for developers, the verified status exactly guarantees higher limits for sending messages in chat or executing commands via IRC-server. So, in theory, if a Twitch confirms via this form that this account is being used as a bot, could this protect me from being blocked if the bot responds to a command with a forbidden term because a third party set it up that way?:thinking:

No

You know, I do catch myself thinking I’m worrying too much about this. But it’s not impossible, wouldn’t you agree? Most of the time, it’s like people won’t do that and won’t use forbidden terms as a response to a team and I’m getting worked up for nothing. I’m still worried about it though. I have plans to register a separate account for the bot so mine won’t be affected, but I also don’t want the bot account to be permanently locked and me registering new ones with the endings bot1, bot2, bot3, etc.

So that would be violating the TOS to get around a service ban
And result int them all getting banned at once.