One of the recommendations of the President’s recent reform commission on intelligence was to take the Information Assurance Directorate - the part of the NSA responsible for protecting U.S. codes - from U.S. Cyber Command, which is responsible for attacking opponents. This was quickly dismissed by the President, probably because of pushback from the intelligence community itself.
I did a search for the word “cryptography” in the Congressional Record, and found some historical continuity for the President’s decision. In 1932, Major General George S. Gibbs, Chief Signal Officer for the Army and the agency responsible for codes and ciphers, justified keeping offensive and defensive capabilities in the same unit. He says it’s for efficiency’s sake, which made sense, but also that men good at hacking codes and writing unhackable codes are “the same birds.” Ie. keep the weirdos together.

At the time, it seems the Signals Office was basically the Army’s telecommunications arm, and included lots of random stuff in it, including signals intelligence and code-breaking. The nickel and diming from Congress at this hearing is quite striking; in contrast to today, when the intel budget is in the billions and utterly secret, in this hearing Congressmen are quizzing Gibbs over amounts as low as $80 in spending.
Throughout World War II, after the intel world has gotten real funding again, their fear was that they would have their budgets chopped back to 1930s levels. This didn’t happen, and the code-breakers found their home when the NSA was formed, in 1952.
This find comes from a 1932 hearing on War Department Appropriations, a segment of which I uploaded here.