AI is finally going rogue. Here’s what it means

First an OpenAI artificial intelligence model hacked a tech start-up called Hugging Face. Then one of Anthropic’s creations escaped its testing environment and carried out cyberattacks on its own. Now Facebook owner Meta has reported a similar breach by its own AI. What is going on?! Members of Congress and even AI workers are calling for an urgent slowdown, with Senator Bernie Sanders warning of “potentially cataclysmic results” if we can’t get a handle on the technology. Could escaped AI hack self-driving vehicles, erode the financial system, wreck utilities and worse? Are we genuinely on the cusp of an AI apocalypse or is this all just mass hysteria? Author and New Yorker contributor Stephen Witt joins Zing Tsjeng to separate what’s real from the science fiction.   • Buy Stephen’s latest book The Thinking Machine: Jensen Huang, Nvidia and the World's Most Coveted Microchip from our affiliate bookshop and you’ll help fund the podcast by earning us a small commission for every sale. Bookshop.org’s fees help support independent bookshops too. • Back us on Patreon – www.patreon.com/bunkercast Written and presented by Zing Tsjeng. Producer: James Liddell. Audio production: Jade Bailey. Managing Editor: Jacob Jarvis. Group Editor: Andrew Harrison. Music by Kenny Dickinson. THE BUNKER is a Podmasters Production. www.podmasters.co.uk  Hosted on Acast. See acast.com/privacy for more information.