I believe the testing itself is done in very good faith.

But I believe the team at Antrophic looks for popular use cases like this one to improve their datasets. Same for every other big player in the LLM game.