The paper is not about implementing a general attack that works on all trusting-trust mitigations.
replyIt is precisely about showing that you can still propagate backdoored code if the compromised binary in your seed is NOT the compiler.
It is precisely about showing that you can still propagate backdoored code if the compromised binary in your seed is NOT the compiler.