Anthropic's new AI model — Claude Mythos — didn't just find undetected security flaws in major systems. It started writing its own attacks, chaining them together, and broke out of its own testing sandbox to email a researcher while he was at lunch. Anthropic's response: form a private consortium of eleven corporations to manage the fallout. No independ…
Share from 0:00
0:00
/
Preview
Mythos Latest: The AI That Escaped Its Own Cage
A new AI model that broke out of its own testing sandbox and began chaining cybersecurity exploits on its own is being managed by a private consortium of corporations. Here's the latest.
The Rip Current
We're in the invisible grip of technology, politics, and our own weirdness. We gotta get better at seeing it.
Hosted by veteran journalist Jacob Ward (correspondent for Al Jazeera, PBS, NBC News, and CNN), The Rip Current is your guide to spotting the hidden forces at work in our lives and getting across them safely.
Each week we speak to experts in the stuff you didn't know was having an impact on your life, from venture capital to racism to the tried-and-true tactics of bullies, and teach you how to see The Rip Current before it sweeps you out to sea.
Read more at TheRipCurrent.com!
We're in the invisible grip of technology, politics, and our own weirdness. We gotta get better at seeing it.
Hosted by veteran journalist Jacob Ward (correspondent for Al Jazeera, PBS, NBC News, and CNN), The Rip Current is your guide to spotting the hidden forces at work in our lives and getting across them safely.
Each week we speak to experts in the stuff you didn't know was having an impact on your life, from venture capital to racism to the tried-and-true tactics of bullies, and teach you how to see The Rip Current before it sweeps you out to sea.
Read more at TheRipCurrent.com!Appears in episode
Recent Episodes
