If with LLM's you can deanonymize at scale, on a personal level, you should also be able to figure out what posts are leading to this deanonymization and remove them or modify them.