The reality of Internet filtering and firewalls, and a rule generalisable to
any attempt at control and autonomy, is that the effect-to-effort ratio matters. The principle of a small effort with a large result is behind the architecture of every switch, gate, door, valve, or dam.
New generic TLDs have the disadvantage of being recently unleashed. There are no venerable sites on XYZ, or its siblings. Much of what's registered there, and that word was "much" and not "all", is absolutely unworthy crap. And for those who are faced with defending either their own or their customers, clients, users, employees, or other stakeholder's security and time, wholesale blocking of the entire TLD solves a lot of problems with very little downside cost.
The obvious response is "but there's a lot of crap on legacy TLDs as well". Yes, there is, but there are also valued, venerable, and essential domains, and blocking all of them is not a viable option. (Though the prospect of whitelisting is becoming increasingly attractive.)
I've known people who are, on the one hand, Internet freedom advocates of decades-long standing --- before most people reading this were born. Who wholesale block access by all China ASNs to their webservers --- because all they see from such networks is malicious traffic. Again: effect-to-effort ratio here is high.
No, it's not "fair". Yes, there's collateral damage. But you're absolutely fighting not merely human nature but all of control theory in trying to combat this.
Register on XYZ and you'll be increasingly fighting a common practice of default-deny, whitelist-by-request. For every user you're trying to reach.
And you should ask yourself if it's really worth it.
XYZ, meantime, are mining and arbiraging short-term cashflow for long-term reputation at the specific expense of its legitimate customers. Those with the least bit of sense will abandon the registrar, leading to an ever-accelerating reputational death spiral.