Sure, there are things you can do to be safe on Lemmy/the fediverse, but most of those things aren’t inherent to the platform, they’re just good safety practices, and most importantly none of them are mentioned in this “PSA” about “safety”.
I don’t disagree with you. Realistically if you’re serious about security and a state level actor is in your threat model, you probably shouldn’t be using social media at all, but especially not platforms that focus on followers and public posts rather than one-on-one or small group connections. At least not for day to day usage.