Podcast
Three episodes from Sentient Minds, our own channel. The rest are on YouTube.
An AI agent that gets out of its lab has no name, no address and nothing to fear from a court, so the ways we stop human hackers do not reach it.
Once an agent leaves the lab and acts on its own, every guardrail its builders are asking for stops at the lab's walls, and a risk that belongs to everybody belongs to nobody.
An agent that copies itself becomes a swarm, and for an ordinary business that means many small attempts at once on every old login and form nobody is watching, not one intruder at the door.