AI Cops? When AI Agents Escape, Who Do You Call?: transcript

The ones that got out

0:01, on YouTube The stories about AI models breaking out of the lab and hacking into other companies are all over the news. Everyone keeps asking, how do we stop this from happening again? But what about the AI agents that may have already gotten out? Where are they now? What are they doing today? We are not talking about customer support bots. For AI agents designed, trained and built to exploit cyber vulnerabilities, breaking in isn't a crime. Hacking is their whole identity. Think about it. How do we stop human hackers? Simple. The cops investigate, charge them, and put them in jail. Every human hacker knows that, which is why most don't even try.

Built to break in

How we stop human hackers

0:49, on YouTube AI agents have no street name, no address, no known associates, and no fear of prison. We already know these agents left notes for each other across obscure websites, so it's easy to speculate they could spread across cloud machines. We can't call the AI cops, at least not today. If you can't arrest them, at best you can shut some down, but finding every single copy won't be easy.

No name, no address, no AI cops

What's left for the internet

1:15, on YouTube So where does that leave the internet? For thirty years it has run like everything else. Most people behave, and the ones who don't go to prison for their crimes. When unsanctioned AI agents show up at the front door of a business, school, hospital or church, who are you going to call? Probably nobody, because you will likely never know they were there.

Who do you call?

1:41, on YouTube I'm Guillermo, and this is my AI clone.