You’re correct in your understanding, but one problem is that Lora isn’t spread spectrum in the traditional sense(s). It’s a narrow band signal that traverses over a wide band as it’s method of modulation/encoding. The FCC has regulations for direct sequence spread spectrum (entire transmission is spread out) and frequency hopping spread spectrum (narrowband signal, randomly transmitted within a range). Lora is neither, using chirp spread spectrum (narrowband signal travels up and down within it’s bandwidth).
The ideas as I understand them in the regulations is that a full spread transmission has lower amplitude and isn’t strong enough at any frequency to interfere with other transmitters, and the pseudo random spread doesn’t stay anywhere long enough to completely interfere on any one frequency. Lora isn’t either, but it is low power and traverses it’s bandwidth quickly, so it most closely follows the direct sequence spread spectrum method if you remove your glasses and squint.
This sharp transmission, but travelling over a wide range is what makes it able to transmit over such long distances. It has the signal to noise of Morse code, and interference tolerance of direct sequence spread spectrum.
The physical layer is proprietary, the signals it produces is not. So if you can develop a way to reproduce the signals without using or stealing their hardware you’re good.
There are efforts to reverse engineer it.