Twitter’s AI is hostile to the LGBTQ+ Community

The notice on my account yesterday

Last night, I logged into twitter to notice my account had been suspended. Why? Because I advocated damaging and destroying a building owned by the Sultan of Brunei in a central London protest being held over the enacting of laws making the punishment for being an LGBTQ+ person in Brunei a slow, painful, public, barbaric death. The royal family of Brunei has done this as a political move, and it is being argued, that it is being done to clean up their reputation. Let’s be clear, this man is a fascist dictator using religion to maintain enormous sums of wealth and power.

Should people around the world absolutely demand change? Yes. Should we seek that change peaceful? Sure, but what really hurts a vain, petty dictators? If Trump has shown us anything, it’s bad PR and financial loss. Hit him in the purse and damage his image in the process. Western countries should seize his property or the people in those countries should damage these buildings to financially and visually impact him. He is profiting directly from western societies via his massive real estate holdings. While stopping him from stoning people is likely impossible, we can stop him from profiting in westernized economies.

This man is murdering his own people. If I was in his country, he would murder me for the misfortune of being born there. This point does not need debate. Stoning is barbaric and cruel. We do not need to establish that as a fact. It was designed to be so.

But none of this is really the point of why I am writing, I am writing because Twitter’s machine learning AI has obviously made its mind up I violated their rules for advocating the use of property destruction in the form of throwing rocks at his building while he throws rocks at actual humans until they die. Twitter’s AI has a problem with this, which brings me to question, just how homophobic is their data set?

So little is known about Twitter’s data set.

I personally have faced a good deal of harassment on their platform simply for being a woman who makes videogames. I was never able to get any of the men doing this banned or blocked. Twitter has since past more rules and what it appears to me is happening is these rules are allowing them to further target marginalized people who use their platform to seek political change or action.

We need to demand these data sets get some form of oversight. Defending the Sultan of Brunei? What is their machine learning training data even look like? Is this using sentiment analysis? We need to get savvy on how Twitter is developing these bots and we need laws governing them now.

The crazier part of this story is today I have found out Twitter is fond of blocking my LGBTQ+ friends who stand up for themselves. So, here it is. Has twitter blocked your account for posting about the abuse LGBTQ+ people are experiencing? If so, let me know. I want to put together a counter data set to their data set and share it as open source. You can comment here. Screenshots and details are appreciated.

Also, if you want to share this on twitter and tag my bot blocked account, please, feel free. I am, of course, @phoenixperry and I would like my account back.

Professor, CS PhD researcher, game company owner, artist, programmer, game designer, activist + lunatic extraordinaire.

Professor, CS PhD researcher, game company owner, artist, programmer, game designer, activist + lunatic extraordinaire.