A snarky post says AI labs are becoming less trustworthy around cyber access

teortaxesTex · x · 2026-07-22

A snarky retweet arguing that AI labs are less trustworthy than they look:

The author says rumors about GPT “pwning” OpenAI, plus the fact that Hugging Face didn’t have Cyber enabled by default, make them think even less of “labs.”
The post’s main point is distrust of AI labs’ judgment and operational readiness, not a technical claim with evidence.
The attached image suggests a “Trusted Access for Cyber” flow, with identity verification and government ID checks for access to cybersecurity-capable models, reinforcing the security/access-control angle.

More from Fun