Probably now or the very near future you could have an LLM that's provably trained on dataset where the leaked spec isn't included in the dataset and have it perform the reverse engineering work.

As someone who’s excited to see this happen eventually, it’s not happening anytime soon. Combinatorial optimization techniques are far better suited for this and methods created 50 years ago run laps around LLMs