I want to ask this question to people who're deeply involved in reverse engineering. Do you think current level of AI models can pull this off ?