- Twitter/Emily Crose
- In 2016, researchers found more than 900 hate groups operating in the US, with over 100 of them having a presence on Twitter.
- Twitter has declared a ban on them and last month the purge began. But Twitter also got heat for doing it wrong when it banned a satire site.
- Ex-NSA analyst Emily Crose has been single-handedly trying to make this problem easier to solve by developing an AI system that identifies “dog whistle” images.
Twitter and Facebook say hate speech is a violation of their policies but they also say it can be hard to identify who is engaged in bona fide hate speech and who isn’t.
Twitter demonstrated the problem earlier this week when it came under fire for blocking a German satirical magazine’s Twitter account after it parodied anti-Muslim comments.
Enter Emily Crose, a former NSA analyst, cybersecurity professional and former Reddit moderator.
Crose was appalled by the neo nazi marchers at Charlottesville, Virginia, and the violence that killed three people and injured dozens more. One of the people attacked and injured was Crose’s friend.
So she’s taken it upon herself to develop a technology she calls Nemisis. It uses artificial intelligence to identify white nationalists by scanning the pictures and videos they post. Nemisis is looking for the symbols they use to identify themselves, their so called “dog whistles.” These might include neo-Confederate flags or certain takes on Pepe the frog.
Her goal with NEMESIS is to help educate people but also to show social media sites how they can find white supremacists and “enforce some decency on their own platforms,” she told Motherboard’s Lorenzo Franceschi-Bicchierai.
Crose explained to Business Insider that the idea for the tech didn’t come directly from her work at the NSA. “What I did for Nemesis I had to learn entirely on my own.”
She used TensorFlow to train her app to spot the images, a popular free and open source AI development technology created by Google.
“I’m currently working on porting the beta system to a production system that we can then run on a front end with Twitter,” she tells us.
In other words, she’s just about ready to use her homemade tool to go hunting for white supremacists on Twitter.
“If I can do this as an individual with few resources, @TwitterSafety should be able to do this easily. The question is, why haven’t they?” she said in a tweet.
Last month, Twitter proved that it was getting serious about the problem, although it remains to be seen how well it will be able to root out hate groups and keep them from coming back. Twitter did begin suspending accounts on December 18, dubbed the #twitterpurge.
Perhaps Nemesis will inspire Twitter to do more. Here’s a peek at how it works from a video Crose posted on Twitter.
— ð~ð»ð²ð°ð²ð·ðªðµ ðx¾ ð¢ðð· (@emilymaxima)December 2, 2017