Tech News Weekly 456

Dan Moren of Six Colors joins the show this week! Meta's Muse has a zero-day vulnerability. Rat brains can power an AI model? Why are we hearing more about AI agents escaping their testing environments? And Meta's new VR glasses.

• Dan talks about a new zero-day exploit that affects Meta's Muse AI agent, which allows an app or terminal command that can impersonate the AI agent and easily gain access to users' messages, emails, and other sensitive data.
• Mikah is amazed at the Biological Computing Company's new AI video model that was built from patterns utilizing lab-grown rat neurons.
• Mikah walks through The Verge's report on why researchers can't simply air-gap AI agents that keep escaping test environments and hitting real-world targets.
• And Scott Stein of CNET joins the show live from Meta's Headquarters to share his hands-on impressions of Meta's new VR glasses that were unveiled at this year's Meta Connect event.

Download or subscribe to Tech News Weekly at https://twit.tv/shows/tech-news-weekly.

Join Club TWiT for Ad-Free Podcasts!
Support what you love and get ad-free audio and video feeds, a members-only Discord, and exclusive content. Join today: https://twit.tv/clubtwit

Links

• Muse, Meta's extraordinarily privileged AI assistant, has a serious 0-day - Ars Technica
• Meta’s New Muse AI Agent Read My Private Messages. I Never Asked It To
• AI Models Built From Rat Brains Just Got Closer to Reality | WIRED
• Why can’t we just keep rogue AIs off the internet? | The Verge
• So Many New Meta Glasses: Muse AI Everywhere, and Ray-Bans Without a Camera