If their system allows bad actors to apply bias, that's literally a systematic bias. Unless they're claiming all of their 'bad actors' are symmetrically distributed along every relevant axis[0], that's the same thing as saying that there is a systematic bias.
There is no reason to believe that any 'bad actors' are perfectly representative of the population at large and cancel each other out, and plenty of reasons to suspect otherwise, even based solely on our prior beliefs about Facebook employees (before this story broke).
[0] I say 'every' axis, because even though we're talking about a one-dimensional spectrum (liberal/conservative), in reality, there are many different dimensions along which bias could be introduced.