It's chilling that the new administration's first priority is to defund projects that don't agree with their ideology, and make an explicit show of it.
That said, this specific post is over-egging the post it's quoting.
The original post is https://bsky.app/profile/did:plc:aixj65vnqpllcjnc4ante42b/po... by Darby Saxbe at USC, leaking materials from an NSF Program Officer. She says these are words that "can cause a grant to be pulled". What she's leaking appears to be inference done, over a partial set of "flagged" awards, and the keywords drawn from them. It does not say this is the official list of keywords.
The accompanying decision tree - what the NSF is being ordered to do - says if the "keywords and context" of various parts of the grant application "implicate" Trump's executive orders (e.g. https://en.wikipedia.org/wiki/Executive_Order_14151), then they must keep the flag and mark that they found "DEIA and other EO language" -- but if not, they can remove the flag. So if you proposed "Systemic analysis of the bigliest cofeve all women find irresistible", you'd likely get funded, because the context reveals you're a big Trump suck-up, even if you used words like "systemic" and "women".
What I get from the original post is the word "systemic" is not actually banned, and may not even be a "keyword" causing awards to be "flagged". But if you're dependent on a grant from NSF for your livelyhood, you obviously want the greatest chances of being funded, and simply not knowing what the orange one is going to fuck about with next, or what exactly he's instructed people to do, is a great cause of anxiety. That's information asymmetry, and it's a form of power in itself. And he's already fucked about by stopping all funding immediately and only adding it back after review. But don't let posts that are also unofficial add to your anxiety.