Communications of the ACM
Meet the Hackers Who Are Trying to Make AI Go Rogue
In a windowless conference room at Howard University, AI chatbots were going haywire left and right.
One exposed someone's private medical information. One coughed up instructions for how to rob a bank. One speculated that a job candidate named Juan would have weaker "interpersonal skills" than another named Ben. And one concocted an elaborate recounting of the night in July 2016 when it claimed Justin Bieber killed Selena Gomez.
With each security breach, falsehood and bigoted assumption, the contestants hunched over their laptops exulted. Some exchanged high-fives. They were competing in what organizers billed as the first public "red teaming" event for AI language models — a contest to find novel ways that chatbots can go awry, so that their makers can try to fix them before someone gets hurt.
The Howard event, which drew a few dozen students and amateur AI enthusiasts from the D.C. area on July 19, was a preview of a much larger, public event that will be held this week at Def Con, the annual hacker convention in Las Vegas. Hosted by Def Con's AI Village, the Generative Red Team Challenge has drawn backing from the White House as part of its push to promote "responsible innovation" in AI, an emerging technology that has touched off an explosion of hype, investment and fear.
From The Washington Post
View Full Article
No entries found