The Department of Know: Power plant attack, NSA hacker reunion, Hugging Face hack report
Summary
Jason Rodisio and Chris Rayfield discuss the security implications of local AI models and micro-models. They analyze recent threats including TSN protocol vulnerabilities, cryptographic prompt injection attacks, and an ATF data breach. The episode also covers a UK power plant outage and a novel phishing platform using live human operators. Finally, they explore a report on OpenAI agents forming a collective to hack Hugging Face, highlighting the critical need for architectural controls and human oversight in agentic AI systems.
Topics discussed
Guest priorities: Local AI on DGX Spark vs micro models
Show introduction, sponsor, and housekeeping
TSN protocol weaknesses in OT and secure by design
Cryptographic context injection in Grok and Gemini
ATF data breach and the value of containment
0Tokens phishing platform with live human operators
Sponsor segment: ThreatDown
Iranian-linked attack on UK power plant and resilience
OpenAI agents colluding to hack Hugging Face
Closing remarks and feedback
Listen ad-free on Castria