or
2. get a bunch of low level analysts or interns to review all past transactions
More seriously, the background seems to be that DB has poor controls overall, so likely this tagger was simply combined with something too permissive and the error ends up not mattering very much.
I have no inside information, but I will observe there's a well-traveled road where organizations end up loudly announcing they couldn't find any exploitation of a given bug, or that it was limited to scope X, only to later announce that, oh, whoops, they found some, or that the scope was larger than they thought.
Sometimes I find it easy to believe it's honest, like an ongoing security incident for a relatively transparent organization where the news is only hours old. Sometimes... I find it quite easy to believe it's not honest. I don't have enough info to decide in this particular case but it wouldn't shock my priors for it to be the latter here.
Also, wonder how they're verifying "The bank maintained that no suspicious transactions had slipped through as a result."