@DaveMWilburn@infosec.exchange
Post #4140009
2026-07-27 15:13 UTC
@mttaggart@infosec.exchange
IMO there's no reasonable way to vet and manage access control to models without guardrails for a global customer base of this size and with so many conflicting interests. And some of the GenAI firms like Anthropic are going to be reluctant to remove all guardrails for their customers (e.g., their conflict with Trump/Hegseth).
Replies (1)
-
@mttaggart@infosec.exchange 2026-07-27 15:21
@DaveMWilburn@infosec.exchange I would push back on this. There is absolutely a way to build the access control you're describing; that part is not a mystery. But what I think you might be driving at is that nobody wants to accept the downside risk. And that's right; the technology is inherently so untrustworthy and unreliable, yet so dangerous, that it defies traditional guarantees of liability.