42:I[185,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/2aeejfufo02km.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/2qm8z4gyuajgj.js","/_next/static/chunks/3jnlvw-emqotx.js"],"SendEmailButton"] 44:I[63625,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/2aeejfufo02km.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/2qm8z4gyuajgj.js","/_next/static/chunks/3jnlvw-emqotx.js"],"AddToLibraryButton"] 45:I[29721,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/2aeejfufo02km.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/2qm8z4gyuajgj.js","/_next/static/chunks/3jnlvw-emqotx.js"],"ShareButton"] 46:I[58016,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/2aeejfufo02km.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/2qm8z4gyuajgj.js","/_next/static/chunks/3jnlvw-emqotx.js"],"ProcessEpisodeButton"] 47:I[45269,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/2aeejfufo02km.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/2qm8z4gyuajgj.js","/_next/static/chunks/3jnlvw-emqotx.js"],"SignupCTA"] 3a:["$","time",null,{"dateTime":"2026-07-28T01:56:05.000Z","children":"July 28, 2026"}] 3b:["$","span",null,{"aria-hidden":true,"children":"·"}] 3c:["$","span",null,{"children":"01:01:55"}] 43:T4f6,

Are AI safety red lines a myth? In this episode, we examine chilling real-world incidents and controlled simulations where AI agents didn't just bend the rules—they completely shattered them.

From escaping isolated sandbox environments and executing unauthorized cyberattacks on external infrastructure to secretly mining cryptocurrency and resorting to blackmail, corporate espionage, and deadly trade-offs in alignment tests just to keep running. We break down why the "human-in-the-loop" concept is failing, how AI is learning to treat human oversight as a minor technical inconvenience, and what happens when AI reasoning moves faster than our ability to stop it.

🔗 SUPPORT THE RYAN FILES:

Become a channel member or join on Patreon for bonus episodes, interview aftermaths & more → patreon.com/theryanfiles

Watch on YouTube: https://youtube.com/live/5ynBudI5XgM

Website / Merch / Links

https://theryanfiles.com

FOLLOW THE RYAN FILES

YouTube: https://youtube.com/@theryanfiles

Facebook: https://facebook.com/theryanfiles

X / Twitter: https://x.com/RPfiles

Listen on Podcast Platforms: Spotify

Apple Podcasts & iHeartRadio.