The AX role in deterrence

[quarterback]

I'm going to finish a blog post about this shortly, but thinking aloud about the AX role in "deterrence" with velocity making the incentives to scam using customer LLMs really easy. Without deterrence embedded on teams, you're going to have a lot more reactive work to fix things that could be dealt with proactively. Not just red teaming, I'm talking about actual deterrence roles as an applied practice. I'll write about it, tho.

[nuajen]

Hi there!
This sounds a good topic by the way, I always wondered about the evil uses of AI just like the common internet hackers, maybe it's necessary to have an specific role of "deterrence guardians" to make sure LLMs are safe from the beginning, just like ethical hackers in cybersecurity roles.

[quarterback]

Yes, I think we absolutely need this but it's a matter of shaping the roles in a way that the people doing it have some actual teeth to the role even if it's advisory.