A few minutes every morning is all you need.
Stay up to date on the world's Headlines and Human Stories. It's fun, it's factual, it's fluff-free.
Every year, Las Vegas hosts the DEF CON hacking conference. Over 3,000 hackers gather in the city to break into the most advanced cyber systems that exist – all above water. But at this year’s DEF CON conference, things are looking a bit different than usual.
Over the weekend, DEF CON kicked off, and artificial intelligence (AI) was the target of one of the “villages,” or group focuses at the event. The conference is actually backed by the White House and held with the cooperation of leading AI companies. This year’s DEF CON AI Village was probably the largest public security test of large language models so far. Everyone’s eyes were on this section of the conference, as we’re right in the middle of the AI boom, and the public is still hesitant about this technology.
According to technology writer Shannon Bond, the competition is important to the AI companies participating because: “they do this kind of testing internally. But, you know, here at DEF CON, it's a much larger set of people from many different backgrounds who are who are testing this stuff out. And they're hoping they're going to find problems that the companies aren't even aware of. And then they'll use that data to help improve their system.”
Basically, the hackers are working their magic to encourage these systems to go off-track and reveal info that they’re not supposed to, which would expose structural weaknesses. Whatever findings are discovered by the hackers will go into reports on AI vulnerabilities to be published next year. Many people are hoping that this hacking challenge will provide and encourage more transparency in the AI sector of tech.
"One of the challenges with this technology being so expensive to produce at the frontier is that it means that unfortunately, a lot of the knowledge and experience with these models is locked up within a small number of well-funded private companies," Michael Sellitto, head of geopolitics and security at Anthropic (an AI safety and research firm) explained to Axios. "The organizers for the challenge are bringing in a really diverse group of people that are not the kind of normal people who work on the technology.”