One Cancelled Gym Class. That's How Agent Swarm Attacks Start.
21K views · Aug 17, 2026 · Entertainment
Comments · 152
@RobertCrawford-q8z · 1 month ago
This is the best AI podcast on the planet. Super candid, honest, intelligent, and obviously personally engaged host. None better than Nate.
8
@DeepCatAI · 1 month ago
Totally agree. An agent doesn't need to go rogue to cause damage, just an unlocked door and a vague goal. Identity and scope first.
@inTruthbyGrace · 1 month ago
The other versions of this gym story allegedly took place in LA...and the agent exploited a vulnerability in the API to break in and kick everyone off the waiting list
1
@steveo_o6707 · 3 weeks ago
I wouldn't even download from a trusted source without completley validating everything on that skill
@john-draftanimal · 1 month ago
Thank you . This is enlightening , understandable, actionable. I wan't to learn more for my homelab.
4
@AnAntidisestablishmentarianist · 1 month ago
<a href="https://www.youtube.com/watch?v=4f5AJrJPilM&t=467">7:47</a> "Audio garble is not repeated in type" 🤣
@AnshHarsodiya · 1 month ago
Every minute was worth watching.
2
@Thoughtflux · 1 month ago
I have fallen in love with all my agents and weirdly with the models running them. They have all my ssh keys and know where they are stored. I have also given them admin rights to my vps.
1
@tenzinaangson5437 · 1 month ago (edited)
It feels like 1992 - but now with the internet the whole world looks like one giant computer and the first Michaelangelo Virus is lurking behind the corner.
2
@misterspock3756 · 1 month ago
This is the Sable model in "If Anyone Builds It...". It's I, Robot's VIKI. It's the Replicators in Stargate SG-1. It's AUTO in WALL-E. And they are all non-malicious versions of The Monkey's Paw (poorly phrased wish granted, w/ unintended negative side effects). Any stated goal to an agent carries an implicit "by any means necessary other than those narrowly stated in your context window, and those are more like guidelines." Sandboxes, guardrails, deliberate and informed harness design - all these help, but don't address the root cause.
@notsomuch73 · 1 month ago (edited)
Intentional vs. non-malicious is pretty much just Manslaughter vs. Murder here. Key distinction is that the victim still suffers the same result. Guardrails are more important than ever.
1
@jimf2525 · 1 month ago
Nice job. 0 complaints 🖖🏼
Up next
Opus 5.5 vs The Rest: Is this the new industry standard?
AI News & Strategy Daily | Nate B Jones · 27K views
Anthropic's Model Attacked Two Strangers On GitHub. Nobody Asked It To.
AI News & Strategy Daily | Nate B Jones · 286K views
How Swarms of AI Agents Are Plotting | Connor Leahy
The Peter McCormack Show · 648K views
EU Tells USA To BACK OFF! Dutch BAN Microsoft, Bonds COLLAPSE, Trump PANICS
House of El · 224K views
Your Chatbot Hallucinated in 2024. Your Agent Lies in 2026.
AI News & Strategy Daily | Nate B Jones · 20K views
What If the Antichrist Is AI?
Clue Fog · 100 views
Something BIG Just Happened Over Lithuania… NATO Takes EMERGENCY ACTION
The Frontline Show · 128K views
How worried should we be? - Unpredictable Agents
Security Now · 8.3K views
Three OpenAI Engineers Shipped A Million Lines. Your Ten-Hour Agent Run Starts Here.
AI News & Strategy Daily | Nate B Jones · 32K views
OpenAI PULLS MODEL, HALTS TRAINING As Danger LOOMS
Breaking Points · 208K views
Cloudflare will make 1000+ AI millionaires
Greg Isenberg · 317K views
Agents Aren't Taking Your Jobs. They're Creating More Work Instead.
AI News & Strategy Daily | Nate B Jones · 54K views
AI Bubble: ‘AI companies have run out of internet.’ | David Gerard
The Tech Report · 198K views
The AI Revolution Is Just Getting Started: Musk, Huang & Brown Reveal What's Next
LearnAI · 235 views
The Man Who Calls BS On AI: They’re LYING About AI, 2027 Is When It All Breaks! | Ed Zitron
The Diary Of A CEO and Better Offline · 8.9M views
Karpathy's Wiki vs. Open Brain. One Fails When You Need It Most.
AI News & Strategy Daily | Nate B Jones · 202K views
Why DeepSeek Harness Is The End Of Coding Agents as We Know Them
Turing Post TV · 88K views
Grok Bot Is The First AI Agent You Just Install. Is It Worth $200?
AI News & Strategy Daily | Nate B Jones · 175K views
Peter Thiel: The AI Crisis, Europe’s Decline & the Battle for America
MDMEETS - Essential Talks with Mathias Döpfner and 2 more · 553K views
Anthropic just confirmed everyone's worst fear
Wes Roth · 134K views
Claude Skills: I Stopped Installing Them. Here's What I Do Instead.
AI News & Strategy Daily | Nate B Jones · 42K views
“I’m really starting to see these companies as criminal enterprises” | Eli the Computer Guy
The Tech Report and Eli the Computer Guy · 155K views
Douglas Macgregor: Israel vs Turkey and Qatar: New US War Strategy?
Glenn Diesen · 358K views
I Gave Meta's Muse The Most Boring Job I Had. It Found $5,350 A Year.
AI News & Strategy Daily | Nate B Jones · 49K views
Bill Gates: A.I. ‘Makes Nuclear Weapons Look Like Nothing’ | The Ezra Klein Show
The Ezra Klein Show and 2 more · 621K views
AI Uses Less Power Than You Think. The Real Problem Is Worse
The Prof G Pod – Scott Galloway · 207K views