I'm speculating that Mythos access might have been given to the intelligence community, or the United States Cyber Command or a similar organization to allow them to find vulnerabilities and set up shop in the systems associated with anyone who might launch cyber attacks on the US as a sort of mutually assured destruction insurance policy.
It's helpful if adversarial actors believe this could be true, whether it is or not.
My personal views. Empirically, the US intelligence community does get their hands on what they need from tech companies. Some anonymous-sourced media reporting indicates that the NSA has access, and that seems very plausible. And yes, they will use it to set up shop wherever they can. That being said, MAD has never really applied to cyber. Ben Buchanan has a whole book about it basically (The Hacker and The State), where he says repeatedly that cyber is great at **shaping** things (basically carrying out attacks) but bad at **signalling** (which is a necessary component of MAD). But that claim may need to be revisited for the AI age, where we may get a fairly good estimate of a country's cyber capabilities based on the models they have access to.
I'm speculating that Mythos access might have been given to the intelligence community, or the United States Cyber Command or a similar organization to allow them to find vulnerabilities and set up shop in the systems associated with anyone who might launch cyber attacks on the US as a sort of mutually assured destruction insurance policy.
It's helpful if adversarial actors believe this could be true, whether it is or not.
Any thoughts on this?
My personal views. Empirically, the US intelligence community does get their hands on what they need from tech companies. Some anonymous-sourced media reporting indicates that the NSA has access, and that seems very plausible. And yes, they will use it to set up shop wherever they can. That being said, MAD has never really applied to cyber. Ben Buchanan has a whole book about it basically (The Hacker and The State), where he says repeatedly that cyber is great at **shaping** things (basically carrying out attacks) but bad at **signalling** (which is a necessary component of MAD). But that claim may need to be revisited for the AI age, where we may get a fairly good estimate of a country's cyber capabilities based on the models they have access to.