I agree with the other responses you have here.

I'll add one observation: the more petty the censorship, the bigger the backlash. People in general loathe that kind of behavior, and internet geeks with a vivid collective memory of being bullied by people like this? Oh, you reap the whirlwind in such cases.

The same way that we deal with every other thing that social media platforms restrict. You make it against the rules and then enforce them. If you’re feeling fancy you could program some heuristics to determine the likelihood of a message containing deadnames - for example, maintain a list of common targets, look for people tagging them or related hashtags in tweets containing their dead name, and use sentiment analysis, to determine whether those messages should be subject to moderator approval before appearing publicly.

Might not be popular but you dont want to restrict that type of stuff in tech, because people very quickly and easily find ways around it, usually by using new similar words that call back to the prohibited ones.

This is an area where moderation is key, and I think people might need to reconsider who they allow to send them direct messages. Especially I expect there to be a better way to vet someone who is trying to directly reach a stranger.

If you are literally going to open your door to the whole world some bad stuff will fall in, but you dont have to let them into your private areas just because they made it in the front door.

Trust networks are another idea, essentially verifying new people through acquaintances.