I did think of that :-) ... and I explained how I countered that in the paper and during my public defense.

First, the proofs were verified by a separate prover that was itself independently formally verified.

Second, the proofs were manually verified by multiple people. Once you know how to read first order logic notation (which is easier to learn than most programming languages), it's not hard to verify the steps by hand. The paper walks through the key parts.

For more details see the paper :-).