The problem is, not all personal data corresponds to the intuitive notion of "private informations". For instance, I, as a U.S. citizen, would be violating the GDPR if I operated a dumb HTTP server that stores request logs indefinitely and does no other processing, such as "python3 -m http.server". (IP addresses are personal data, and U.S. authorities can make me turn over my logs; thus, I cannot store the logs for however long I want.)
Even when you have clear legal obligations to collect a bunch of private data, you still have to figure what reasonable measures to protect it are. What's a reasonable measure when you're obligated to keep seven years of financial records about your customers in a way that lets you produce ready reports for regulators? Complete with mandatory personal identifiers for tax purposes?
GDPR is not just about being obligated to ask for consent. Its requirements go a great deal further. Probably. Maybe. Depending on what someone unknown with an unknown background considers reasonable based on unknown factors.
I'm sure that will be easy to write requirements around. Right?
It is really not hard at all to comply with GDPR. Don't collect private informations you don't need, and don't share them.
Now what's hard is to build a system that complies with GDPR's letter _and_ breaks GDPR's intent. And you know what? That's intended.