Given AI's strong cybersecurity capabilities, it seems that there is a realistic future where AI will start finding a foothold on all internet-connected devices.
How do we continue to share information between humans, while AI is wreaking havoc on the systems that used to provide the means to do so?
Will we be airgapping our personal devices and using known-infected ones to access the internet?
what do you think you will lose if AI hack your personal device vs. the databases that holds its data at third parties.
I do think things like maybe some nas gateway might be interesting if u have things like a NAS.
idea would be that through some proxy u connect to it, so on that proxy u can set allowed times and detect anomalous access pattern.
the proxy would have 2 nic. 1 to network and 1 with NAS.
why? because the proxy u can have an agent write in 2 prompts, and is unknown layer for an attacker.
it works rly well and is super easy to do.
i cant think of any other personal device which would merit protection like that unless you have some PC that similarly serves as local network data storage.
In all cases its a good idea to have as little as possible open to the internet. the problem often is that things like home routers will actually have ISP ports open that are attackable, and for those types of things ur really at the mercy of your ISP. (modem/router mgmt ports etc.) This can vary _wildly_ between ISP.
So then u revert back to either keep ot offline or have _custom_ protective layer around it.
cant stress enough that _custom_ is very important today.
open source or proprietary cyber solutions are under large fire. unless u are really diligent maintaining it and patching, its likely not gonna hold forever.
a custom solution to apply simple ACL and anomaly detections on access patterns is really hard to bypass. atleast u can get yourself a notification.
for myself, i think its too early for this. there is not much more amplified attack on home systems as there already were. first hit will be SMB, maybe enterprises, and then individuals. usually. (bigger players develop the tools n methods and they dont target individuals mostly. skids and psychos pick those tools up and point em at grandma...)
I agree with nouridin: Its probably the fear that AI would attack lethal entities that drive the AI tech-bros to slow down the development of better AU models.
AI has no autonomous goals or personal ambition. The only way they would "take over" in any sense is if people give them a goal that drives them to that direction and give them access to freely access any system.
What needs to stop is people doing exactly that. The setup of, "Look, I gave my AI a goal and system access!" is exactly the wrong thing to do. As we've seen already with the various anecdotes of people's LLMs destroying their data.
There are a plethora of tools and systems in this world that can do great harm if used improperly. AI is now one of them. We need people to start thinking about them in that context. "This is a tool. There are helpful ways to use it and harmful ways to use it. I should take care to stay on the helpful side."
For strictly personal stuff, like my home network, isolate behind a router to block all outside traffic. Then install a VPN on your home server so that you can remote in via a tunnel. I use TailScale's free tier and its great.
If that is not good enough then do the same as above but lock everything behind PKI certificate access and make all your home services accessible only via mTLS. This is huge pain in the ass, but you are that paranoid you get what you pay for in your time and effort.
What happens when a nefarious cabal trains an AI that can command every internet connected device to "del *.*"?
All our texts, photos, documents, databases, banks,..., gone, unless backed up to air-gapped data centers.
How do we continue to share information between humans, while AI is wreaking havoc on the systems that used to provide the means to do so? Will we be airgapping our personal devices and using known-infected ones to access the internet?
what do you think you will lose if AI hack your personal device vs. the databases that holds its data at third parties.
I do think things like maybe some nas gateway might be interesting if u have things like a NAS.
idea would be that through some proxy u connect to it, so on that proxy u can set allowed times and detect anomalous access pattern.
the proxy would have 2 nic. 1 to network and 1 with NAS.
why? because the proxy u can have an agent write in 2 prompts, and is unknown layer for an attacker.
it works rly well and is super easy to do.
i cant think of any other personal device which would merit protection like that unless you have some PC that similarly serves as local network data storage.
In all cases its a good idea to have as little as possible open to the internet. the problem often is that things like home routers will actually have ISP ports open that are attackable, and for those types of things ur really at the mercy of your ISP. (modem/router mgmt ports etc.) This can vary _wildly_ between ISP.
So then u revert back to either keep ot offline or have _custom_ protective layer around it.
cant stress enough that _custom_ is very important today.
open source or proprietary cyber solutions are under large fire. unless u are really diligent maintaining it and patching, its likely not gonna hold forever.
a custom solution to apply simple ACL and anomaly detections on access patterns is really hard to bypass. atleast u can get yourself a notification.
for myself, i think its too early for this. there is not much more amplified attack on home systems as there already were. first hit will be SMB, maybe enterprises, and then individuals. usually. (bigger players develop the tools n methods and they dont target individuals mostly. skids and psychos pick those tools up and point em at grandma...)
to be honest i am afraid of AI hacking lethal entities like nuclear power stations more than hacking my personal phone or pc
I agree with nouridin: Its probably the fear that AI would attack lethal entities that drive the AI tech-bros to slow down the development of better AU models.
AI has no autonomous goals or personal ambition. The only way they would "take over" in any sense is if people give them a goal that drives them to that direction and give them access to freely access any system.
What needs to stop is people doing exactly that. The setup of, "Look, I gave my AI a goal and system access!" is exactly the wrong thing to do. As we've seen already with the various anecdotes of people's LLMs destroying their data.
There are a plethora of tools and systems in this world that can do great harm if used improperly. AI is now one of them. We need people to start thinking about them in that context. "This is a tool. There are helpful ways to use it and harmful ways to use it. I should take care to stay on the helpful side."
I don’t think so. But proper firewall setup, DMZs and guest networks, patched devices (so open source devices…) go a long way…
For strictly personal stuff, like my home network, isolate behind a router to block all outside traffic. Then install a VPN on your home server so that you can remote in via a tunnel. I use TailScale's free tier and its great.
If that is not good enough then do the same as above but lock everything behind PKI certificate access and make all your home services accessible only via mTLS. This is huge pain in the ass, but you are that paranoid you get what you pay for in your time and effort.
What happens when a nefarious cabal trains an AI that can command every internet connected device to "del *.*"? All our texts, photos, documents, databases, banks,..., gone, unless backed up to air-gapped data centers.