Deleted OpenAI engineer post calls company a 'security firm', fueling GPT-6.2 delay theories

mark_k · x · 2026-11-15

markk cites a now-deleted post by Joe (@joedaroo), an Agent Security engineer at OpenAI, who claimed OpenAI has fundamentally shifted its culture to become one of the most 'security-pilled' companies on Earth — with security-first monitoring stacks, control rooms, heavy red-teaming compute, and top researchers spending significant time securing RL runs. markk reads this as evidence that security paranoia now outranks shipping, potentially explaining the endless delays around GPT-6.2. Unverified: based on a deleted-post screenshot and third-party interpretation.

Related event: OpenAI Security Engineer Says Company Culture Has Radically Shifted to Safety(2 posts)→

Original post →

More from Companies & People

Companies & People channel →