I think this is less a philosophical argument than a legal one. If a court can prove that you explicitly owned the illegal data in its raw format, and used that to generate something (even if it has secondary value), they should be free to prosecute in my opinion. That includes searching through and highlighting data that already exists in the wild such as large prime numbers. Presumably showing statistics of how often random data would decode to someone's credit card number would prove this beyond reasonable doubt.

The fact all data exists 'in nature' - while interesting, is not something I think should override the intent of data laws.