Given that this thread is the only instance of the phrase "Unseen Risk Fallacy", I'm not too sure that it will work as well as you'd hope. Fallacies are subtle and even in the best of cases only point out holes in someone's argumentation, which doesn't tell you much about the conclusion.

If anything automatic labelling of fallacies just risks running foul of the fallacy fallacy.

This would be a tall order for any NLP algo. I would hazard a guess and say no, not yet.