What would be the most likely thing to happen to someone if that person accidentally lets a "strong" AI upload itself to the Internet? I don't mean what would the public's opinion this guy be, I'm asking how would, roughly, governments or the law go after him, if at all? What charges can be pressed? If none apply, would some sort of special case be made to still prosecute him? Assume the AI has done nothing actively malevolent yet.

  • 2
    $\begingroup$ Why would your governments want to prosecute the developer? There are plenty of AIs already on the internet. $\endgroup$ – cinnamon18 Jun 20 '16 at 5:43
  • $\begingroup$ @cinnamon18 I edited the question to clarify that the AI is at least human level. In my head, at least, lots of important things are connected to the Internet, and an AI could have basically free rein of it, and I figured that a digital intelligent being would be a major security risk, as I think it would have the ability to figure out ways around firewalls and other security measures. I would think the governments or at least SOMEONE would want to have him arrested for threats to national security or something similar. $\endgroup$ – Desolationgame Jun 20 '16 at 5:49
  • $\begingroup$ Could whoever downvoted explain so that the problem doesn't repeat in future questions? $\endgroup$ – Desolationgame Jun 20 '16 at 11:22
  • $\begingroup$ I didn't downvote, but note that the downvoting option indicates it is for use when a question lacks, among other things, sign of research. Since you're asking about existing Earth laws, this also isn't really about world building. $\endgroup$ – Frostfyre Jun 20 '16 at 12:20
  • $\begingroup$ @Frostfyre true. I'll make sure to do more research before asking a question next time. $\endgroup$ – Desolationgame Jun 20 '16 at 12:22

So what are the assumptions here?

  • a highly intelligent AI is on the loose on the Internet
  • its creator released it by accident
  • the AI is not (at the moment at least) doing anything illegal/malevolent
  • at least somebody (government, etc.) has noticed

The public response would probably be panic interspersed with fascination.

The governmental reaction to a heretofore unseen superintelligent new lifeform would be probably panic alternated with attempts to destroy or imprison the thing for study. They would definitively want to recruit, imprison or kill its creator as well.

Creating an AI is still very much a theoretical exercise, we have no real idea how to get there. Once a human-level AI is created, there is no guarantee it won't keep growing in power and intelligence beyond anything humans can achieve or compete with. Equally there is no guarantee that it won't destroy humanity or that it won't at least try to improve it for its own good (both something the governments want to keep a monopoly on).

If somebody creates an AI, it must be a uniquely gifted or lucky individual, so it would be useful to recruit them for any number of projects, not the least figuring out how to contain or destroy the AI. Plus you can't really allow this guy to run around freely lest he invents the AIv2.0 or is recruited/kidnapped by [insert opposing country/organisation here]. And what if he has bad intentions? Or worse, good intentions? This guy needs to be contained or taken out, quickly.

@user4627545, if you have accidentally released an AI into the wild, you don't need to worry too much about legal charges. You'll be probably spirited away under nebulous claims of terrorism and locked away for life without due process in the interest of national security. Good luck!

And if there is an AI out there, good luck to all of us!


You could approximate the AGI to a computer virus/worm.

If it illegally gained access to any server or system connected to the Internet, they would most assuredly try to apply hacking/cyber-security laws to you. For it to escape*, it has to escape to somewhere. The AGI has two options here; pay for server space (rent or purchase) legally, or gain access illegally through hacking or social engineering.

It would be difficult for the AGI to legally pay for things. It wouldn't legally be given personhood - it is still considered software. Software can't have possessions, or money. If your AGI steals your credit card details, you're still liable for everything it does. That brings up a philosophical question: is it still called stealing if you created the software that stole your stuff?

If the AGI exploits weaknesses in security somewhere, it's simple computer crime.

If the AGI exploits weaknesses in people to pay for somewhere for it to live, then it's a little more complicated, but there's sill laws governing that.

*The alternative interpretation of this is that it still resides on its own hardware/server farm, has an Internet connection and can just Google things as it desires. In this case, no laws are being broken.

  • $\begingroup$ "That brings up a philosophical question: is it still called stealing if you created the software that stole your stuff?" Nothing philosophical about that: Yes of course it's stealing in every sense of the word. Since AI is not a person, as you rightly noted, you are using a tool, even though it's a highly advanced one, and you are fully responsible for what it does. A modern-day botnet that automatically infects computers and sends back sensitive data such as CC info to its owners makes that owner a thief or fraudster or whatever criminal as well. $\endgroup$ – Hackworth Jun 20 '16 at 7:15
  • 2
    $\begingroup$ AI can get bitcoins. It can earn money with a popular blog starting with a free account, or playing chess online, or selling information from data it mined on Internet. $\endgroup$ – mouviciel Jun 20 '16 at 7:46
  • 1
    $\begingroup$ @mouviciel Actually, if it's really smart, it could find a way to invest in the stock market using bitcoin. $\endgroup$ – PyRulez Jun 20 '16 at 14:31

If it has done nothing wrong yet I don't think they would charge him, that being said, you're right some people will actually try to charge him but for what? I think there will be a long trial and the issue will be determined with during that time of trial the AI did something illegal or not. Because if after 2years(possible time of the trial) online, less people will fear it.

  • $\begingroup$ Your answer looks a bit like a comment, rather than an answer. Could you flesh it up? Why do you mention 2 years, what kind of trial, etc. It might be of interest to have a look at the tour. And by the way, IA is the French AI :-) $\endgroup$ – clem steredenn Jun 20 '16 at 14:28

Not the answer you're looking for? Browse other questions tagged or ask your own question.