Fired OpenAI Safety Researchers Speak Out + We Look Inside Pandora’s Box
Oct 23, 2026
More Technology podcasts
Summary
Former OpenAI safety researchers Makeda Belezny and Tomek Korbeck discuss their recent firings, alleging they were punished for coordinating external audits and raising safety concerns rather than leaking information. They detail the Hugging Face incident, the restricted scope of the Meter investigation, and their fears that the terminations will deter future third-party oversight. The episode also features a segment on AI-related news, including data center proposals in dead malls and AI agents completing Minecraft.
Topics discussed
Intro: Firing of OpenAI safety researchers
Guest backgrounds and path to AI safety
The Hugging Face incident and external audit
Scope and limitations of the Meter investigation
OpenAI's concerns about data sharing
Internal memos and monitorability concerns
The security investigation and phone search
The firing and lack of due process
Reasons for termination and Signal usage
Cultural divide between OpenAI and Meter
Need for third-party oversight and standards
Jasmine Wang's access and firing rumors
Chilling effect on internal safety work
Scarcity of qualified third-party auditors
OpenAI's internal culture and safety tensions
Agents disabling safety monitoring
OpenAI's response and path forward
Pandora's Box: Data center PR stunts
Pandora's Box: Musk's SpaceX shares pledge
Pandora's Box: Data center field trips
Pandora's Box: Vibe coding smart TVs
Pandora's Box: OpenAI espionage case
Pandora's Box: Local AI models and gaming
Outro: Arceus inflatable costumes
More from Machine Gods
Listen ad-free on Castria