Get rid Child predator and Spam Bots
Okay Twitch so here is the Simple Idea - why don't you track chats of new accounts (chatter what they write and new streamer what's in their chat) and games often used by bots and children. e.g. Fortnite, Rocket League, Brawl Stars, Standoff 2, ASMR, just chatting.
Track the words predators and bots use e.g. challenges, Creator code, feet, mouth, ***, "code is", also the common Emotes 🍑🍆💦
I have this Idea because of the massive spam of “use my creator code” and pedophile activities. I’ve seen a boy dancing naked because he thought he got a 1000$ Donation and see kids show feet because they think they get followers. Twitch only react if there is someone to report. If no one reports these accounts are tree to do whatever they want.
My idea in Detail:
Categories:
Check your actually most banned categories where underage reports were done. So these are the mail categories where predators are. For the bots it’s the same group and also most popular games.
Account selection:
Intense watch of accounts younger than 1 Month. Most predators and bots are made one week ago.
Less Intense but still watch chats of live streamer accounts in the selected categories 3-6Month.
You could also Involve just started stream with less than 10 viewers to track them the first 2 hours.
Chat trigger:
Again check your reported accounts for Scam, Bot , sexual stuff and against underage
Check for common words and phrases.
Use a Rating – High Priority and low Priority. In my opinion a new account using one of these words / phrases are inappropriate.
High priority:
e.g. challenges, mooning, show feet, 🍑, 💦,🍆, creator code, /me donated, face reveal, shirt off, add me, snapchat, designer, **, **, follow 4 follow, follow me back
I would instantly ban these new accounts only reviewed if ppl use the tools to unban. Don’t forget the phrases from the reports you see – because also the spam bots change the text like “code is hind” is now something like “get free rare skins use hind”
less priority:
e.g. add me on discord, where do you life, do you have a girlfriend / boyfriend, wisper me
Reviewed if used often. Exclude the known bots like Streamelements Nightbot Moobot ...
You can also think about a kind of “Point System” – If in the chat of new accounts is just the word “foot” “feet” it’s not bad but e.g. someone telling “not to show feet” or “don’t take shirt off”. If ppl use in a new chat 5 or more times they will get reviewed by the support but not instantly banned.
Probably KI can also do this and decide what is a good or a bad phrase. The “Point System” could be good to also filter old accounts who were not identified yet.
I never did anything with Python and I took me only 12 hours to make a bot – that did in parts the same. Filtered a list of live Fortnite streamer and watches the chat about some trigger phrases. Not Perfect bc I would still need to make the list bigger, get the new accounts first and more categories. Also I would still need to report the accounts manually. Especially the bot I don’t care to report them by hand. BUT! You have all the data and also know the new accounts chatting. And you don’t lose time if it is automated.
So PLEASE Twitch – I’m quite sure it’s not that big effort for your Team. You can easily try the system with the massive bot spamings from tion hind itsplunja and streamrise. So you can quickly save the children not to be naked, show feet or open the mouth really wide on Twitch anymore.
Not Chat but also against bots: You could check profile pictures and compare titles of new live channels bc the sex bot streams often use the same picture – links in bio and stream title.