I've been thinking that reverse engineering skills are much more useful than puzzle solving.
I had such a question at Google. I guessed what was the problem they were trying to solve as it was a classic leetcode question (linking nodes in a tree while traversing it). So while the format is a bit different than typical leetcode solving, I'd say it's still the same type of skills and preparation involved (at least in my case).
Also, the bit about puzzles, leetcode is somewhat of a myth, mostly perpetuated by salty candidates who couldn't pass tech interview loops, for whatever reason. Puzzles were explicitly banned at Meta and Google, when I worked there, and if a question is found on leetcode or a similar site (as they eventually do), they'd get banned too. The idea isn't to make candidates jump through a series of pointless hoops, but to concretely evaluate their coding and reasoning abilities. It's certainly not perfect, but it's a lot more BS proof that just having a chat or take home assignments. Also, needless to say, more senior candidates are expected to successfully grapple with harder problems.
Why?
At any rate, there's a chance the accusation was wrong, and if you make wrong accusations and the other person disputes it, it could lead to unnecessary legal proceedings. (libel, etc.)
PS: not saying the company policy by the GP is necessarily correct, just saying what lawyers and risk adverse bureaucrats might be thinking when they drafted it.
Smaller companies follow these trends because they think it will make them successful. It's like reading the latest article on "Ten morning routines of billionaires", following that, and believing that it will make you rich. It's the other way around.