Personally I find this more than a little unnerving. Trying to figure out how to stop AIs from operating from bad intent is no small task, but it seems a lot simpler than stopping AIs that are trying to be “helpful.“
@emollick.bsky.socialThe cybersecurity threat from agents is increasingly more likely to come from a massed swarm of AIs whose only goal is to penetrate your system to figure out how much you paid for your company t-shirts as part of a research effort to "find good t-shirt prices" as it is from bad actor attacks.
