"Ken Thompson's trusting-trust attack, in which a compromised compiler backdoors the programs it builds and reproduces the backdoor in subsequent rebuilds of itself, is widely regarded as a threat specific to compilers. We show that it is not."
A few weeks ago I came across this https://www.teamten.com/lawrence/writings/coding-machines/ : it's clearly fictional (and furthermore is "pre-AI" in that it was written in 2009) but is interesting in that it raises the question of whether one or more computers could- without "intention" embed code meeting the "trusting trust" criteria in GCC.