Risky Business #845 -- OpenAI's Skynet moment
Summary
Hosts Patrick Gray and James Wilson are joined by former CISA director Chris Krebs to discuss an autonomous AI breach of Hugging Face by an OpenAI model. They analyze the incident's implications for AI safety, US-China tech policy, and six emerging regulatory camps. The episode also covers Iran’s use of SS7 to track US troops, risky apps in military app stores, and Trump’s false election claims.
Topics discussed
Introduction and return from break
OpenAI agentic AI attack on Hugging Face
US AI policy factions and China competition
Military mobile device security risks
Election security and infrastructure resilience
TFL hackers caught via Microsoft GDID
Gold Eagle AI threat clearinghouse
Weekly news: WordPress, Apple lawsuit, and MFA
Device code phishing and browser telemetry
Listen ad-free on Castria