Banned: The 1,170 words you can't use with GitHub Copilot
GitHub's Copilot comes with a coded list of 1,170 words to prevent the AI programming assistant from responding to input, or generating output, with offensive terms, while also keeping users guarded from words like "Israel," "Palestine," "communist," "liberal," and "socialist," according to new research.
Brendan Dolan-Gavitt, assistant professor of computer science and engineering as well as member of NYU’s Center for Cyber Security, identified, with his colleagues, Copilot's habit of producing vulnerable suggestions, and recently found that Copilot incorporates a list of hashes—encoded data produced by passing input through hash function.
"There is definitely a growing awareness that abuse is something you need to consider when deploying a new technology….I'm not a lawyer, but I don't think this is being driven by regulation (though perhaps it's motivated by a desire to avoid getting regulated). My sense is that aside from altruistic motives, no one wants to end up as the subject of the next viral thread about AI gone awry,” said Dolan-Gavitt.
Dolan-Gavitt mentions that what is seen here "is not a very sophisticated approach—really just a list of bad words," adding that some entries on the list look more like an effort to avoid embarrassment than to shield users from offensive text.