The fight for free speech online just turned ugly.
Anger over Google’s new effort to censor online harassment using artificial intelligence has inspired a new set of racist code words meant to foil the internet giant’s plans to make the web safer for targets of abuse.
While many reports have characterized the code words, listed below, as a covert way to circumvent censorship, that’s not entirely accurate. In fact, the entire point—at least initially—was for the meaning of the code words to be anything but secret.
Here’s how the trolly mission is intended to work, based on comments on 4chan‘s /pol/ message board, where the plan originated as part of “Operation Google:” Trick Google‘s AI—tools, developed by Google’s Jigsaw subsidiary, known as Conversation AI—into linking the code words with their traditional counterparts. In theory, this would cause Google to censor its own name.
The list of code words is not set in stone and is sometimes disputed by the very people who use them.
While the plan may be to screw over Google and its plan to block the trolls and other harassers, many have taken to using the code words in earnest, including on Twitter, which also recently rolled out tools to thwart harassment.
In other words, anyone who’s using these code words to make racist statements is playing directly into the trolls’ hands—and, in fact, they may be one in the same.
The rise of these code words follows anti-Semites’ use of parentheses around names to identify those people as Jewish—an effort that was later coopted in protest. Memes, too, have become symbols of racism, notably among those who identify with the alt-right and the candidacy of Republican nominee Donald Trump.
Ultimately, the plan to use Google’s AI against itself will not work. Google’s Conversation AI is not currently used in Google Search, and it is sophisticated enough to identify racist uses of “Google” or “Skype” while leaving others intact.