OpenAI Codex sandbox escape, President proposes AI Force, Cyber Command suicides
Summary
Host Steve Prentice covers a range of security news, including a sandbox escape in OpenAI Codex that allows remote code execution and a new AI model from Alibaba that diagnoses abdominal conditions. The episode also discusses Microsoft Teams' updated file extension blocking, congressional efforts to support cyber operators' mental health, and a North Korean campaign stealing $105 million from tech workers. Additionally, Prentice highlights a widespread decoder flaw found using AI, a new method for eavesdropping on headphones via electromagnetic signals, and the US president's proposal to cr…
Topics discussed
OpenAI Codex sandbox escape allows remote code execution
President proposes AI Tzar and AI Force for monitoring
Microsoft Teams adds custom file extension blocking
Congress considers support for Cyber Command personnel
Sponsor segment: Nudge Security AI agent discovery
Alibaba open-sources AI model for medical CT scans
North Korean Water Plum campaign infects 30k devices
AI tools find widespread flaw in software decoder
New method for eavesdropping on headphones via EM signals
Outro and show information
Listen ad-free on Castria