Comment by Jensson

> If feedback from this incident is in its context window, it is highly unlikely to make this same mistake again

If this incident gets into its training data, then its highly likely that it will repeat it again with the same confession since this is a text predictor not a thinker.