Spoofing an AI? Possible?, Matrix mind control? Options

[The Jopp]

Ok, more of a Matrix funny

An AI is a running program similar to a Pilot and Agent – but far more advanced.

They also have a Persona and is a running program on a node.

The AI have it’s own individual firewall apart from it’s improved home node.

Now, it would be difficult, but could you spoof an AI?

1. You observe the AI’s persona
2. You spoof the AI itself and give it an order.
3. Would it believe it was it’s own idea?

[Karoline]

I don't think this would be any more possible than spoofing a technomancer or hacker.

[Aerospider]

The key difference between A.I.s and other autonomous programs is that they are truly self-aware. I would have a hard time buying the notion that the A.I. would send itself commands as a way of taking action - if it needs to command itself to do something, then wouldn't it need to first command itself to command itself? Ad infinitum.

[Karoline]

The key difference between A.I.s and other autonomous programs is that they are truly self-aware. I would have a hard time buying the notion that the A.I. would send itself commands as a way of taking action - if it needs to command itself to do something, then wouldn't it need to first command itself to command itself? Ad infinitum.

Guards! Bring me the forms I need to fill out to have him taken away. (IMG:style_emoticons/default/wink.gif)

[The Jopp]

A spoofed AI should be possible since it by it's core is a software. I would allow it to resist the command very much like mental manipulation spells - makes them a bit more fun as they could be manipulated very much like a human being affected by a spell.

Give them a resist bonus to Mental Attribute+Required Access.

Spoofing Admin access on an AI and giving it a command would be resisted by Mental Attribute (Logic) + Admin Difficulty (6)

[Karoline]

A spoofed AI should be possible since it by it's core is a software. I would allow it to resist the command very much like mental manipulation spells - makes them a bit more fun as they could be manipulated very much like a human being affected by a spell.

Give them a resist bonus to Mental Attribute+Required Access.

Spoofing Admin access on an AI and giving it a command would be resisted by Mental Attribute (Logic) + Admin Difficulty (6)

That makes AI even more absurdly weak than they already are. At absolute best the AI would have 12 dice to resist with, and it isn't even remotely hard to get 18ish dice for an attempt like this. Basically all AIs would be controlled completely by any halfway competent hacker/TM that came along. I mean AIs have a software core about as much as humans do. Sure, they used to be normal software, but we used to be monkeys who used to be fish who used to be single cell organism who used to be atoms. That doesn't mean we're no different from monkeys.

[The Jopp]

That makes AI even more absurdly weak than they already are. At absolute best the AI would have 12 dice to resist with, and it isn't even remotely hard to get 18ish dice for an attempt like this

You misunderstand me. This resistance roll would be made directly after the attacker have managed a spoof command

The net hits the attacker got after the initial opposed test would be pitted against the AI’s Dicepool of 6-12.

Hacking+Spoof-Difficulty VS AI+Firewall = Result VS AI+Difficulty of Command

Since difficulty comes into play Attacker have a -6D6 for Admin test and the AI a bonus of +6 when resisting the actual command.

Since the player itself can rule what and what not is covered under each access level they might be forced to ONLY spoof admin, something that is very difficult in this case.

[Muspellsheimr]

An AI does not accept matrix commands in the traditional sense, so they cannot be spoofed.

Spoofing involves making a command appear to come from an authorized source, and works against Agents & similar software that is designed to accept authorized commands.


A synthetic intelligence accepts commands from an authorized source if it can be convinced to do so, not because it is programed to do so. Thus, "spoofing" an AI falls under Con/Negotiation.

[Karoline]

Now you're just getting convoluted rules that involve the AI getting to make multiple rolls to resist the spoof.

Besides, there is no way an AI would ever accept admin access from anyone. It isn't a node, and so admit access doesn't exist in the first place. And even if it was, there would be no reason that it couldn't just filter out any command as coming from externally, and ignoring it. Most programs can't do that because it only operates based on commands from the outside.

Like I said, this would be just as practical as spoofing a person's brain, because AI are mostly identical to an actual brain.

[The Jopp]

A synthetic intelligence accepts commands from an authorized source if it can be convinced to do so, not because it is programed to do so. Thus, "spoofing" an AI falls under Con/Negotiation.

Good point, I stand corrected.