4 ms·

That's not what I was suggesting at all. If you control the cell infrastructure you absolutely can put cellular capabilities into products that aren't explicit

That's not what I was suggesting at all. If you control the cell infrastructure you absolutely can put cellular capabilities into products that aren't explicitly advertised. That would be a bigger risk for surveillance tech. Your complaint ignores my original statement about controlling both components (cellular and device). There is no need to tamper with someone else's TLS-secured communications or anything, you can just plant cellular capabilities into the devices.

gregmac 7y ago
I didn't pick up that's what you meant by "both of them" so sorry for my misunderstanding. However I still don't think 5G is an important factor here. If you control both ends at a higher layer of the stack (eg, it's your own software connecting to your own servers) you can do whatever you'd like. If you have OTA software updates, you can do whatever you like at any point in the future. If this channel is properly secured with TLS no one can intercept or fake it, and if you want, you can design the devices so it won't work without this connectivity present (see: most cloud-based "smart home" devices on the market today). > If you control the cell infrastructure you absolutely can put cellular capabilities into products that aren't explicitly advertised. That would be a bigger risk for surveillance tech. So many devices already have internet connectivity built-in, and if money is not a concern you can already incorporate a GSM modem and pay for a link to the existing cellular networks -- it's not like AT&T (or whoever) is going to care what you are doing with the connection, so long as you're paying for it. You can also use other short-range radio links. I'll grant you that 5G introduces the possibility to have this link without AT&T's knowledge -- at least assuming they don't have the ability to also detect the traffic coming out the tower's uplink. The big downside of this attack vector, aside from the immense cost and complexity of building this into 5G carrier equipment and all these devices, is if detected and blocked at the uplink all that effort is suddenly for nothing. Basically, the 5G-specific attack vector is expensive and brittle, and thus pretty unlikely. The attack vector at application level is easy and cheap (no hardware-related costs), gets you almost all the same capabilities, and in many cases can be done today on existing in-field products with OTA software updates.

gregmac 7y ago

I didn't pick up that's what you meant by "both of them" so sorry for my misunderstanding. However I still don't think 5G is an important factor here. If you control both ends at a higher layer of the stack (eg, it's your own software connecting to your own servers) you can do whatever you'd like. If you have OTA software updates, you can do whatever you like at any point in the future. If this channel is properly secured with TLS no one can intercept or fake it, and if you want, you can design the devices so it won't work without this connectivity present (see: most cloud-based "smart home" devices on the market today). > If you control the cell infrastructure you absolutely can put cellular capabilities into products that aren't explicitly advertised. That would be a bigger risk for surveillance tech. So many devices already have internet connectivity built-in, and if money is not a concern you can already incorporate a GSM modem and pay for a link to the existing cellular networks -- it's not like AT&T (or whoever) is going to care what you are doing with the connection, so long as you're paying for it. You can also use other short-range radio links. I'll grant you that 5G introduces the possibility to have this link without AT&T's knowledge -- at least assuming they don't have the ability to also detect the traffic coming out the tower's uplink. The big downside of this attack vector, aside from the immense cost and complexity of building this into 5G carrier equipment and all these devices, is if detected and blocked at the uplink all that effort is suddenly for nothing. Basically, the 5G-specific attack vector is expensive and brittle, and thus pretty unlikely. The attack vector at application level is easy and cheap (no hardware-related costs), gets you almost all the same capabilities, and in many cases can be done today on existing in-field products with OTA software updates.