I even have my e-mail signature set to remind people to be careful before opening things in Word; after all, it can be really easy to accidentally betray trust by opening a draft.

(If "consider before printing this e-mail" is a thing, avoiding carbon-chugging LLMs has to be at *least* as important.)

Constructive:

I think the article is perhaps a bit verbose and long-winded to win over a lot of laypeople / disinterested office workers. Adding a "tl;dr: opening documents in Word/GDocs automatically uploads the contents for AI training without the author's consent. Consider alternatives like LibreOffice" might help.

Also, you used the term "LLM" before you use the term "large language model", and the target audience probably doesn't already know what LLM stands for. Same with GAN.

Thanks to everyone who's pointed out that the page is very difficult to read on some systems; I have not had that happen on any of the laptops and/or phones that I have access to, so I wasn't able to notice that earlier, and I apologize for the accessibility issues that creates.

I'm not, unfortunately, a professional webdev, and have found it a little challenging to make a plain HTML page that works everywhere.

Someone in DMs very kindly suggested a fix, and I'll look at applying that ASAP.

Nice, will share!

PS: The link to your name in the footer is broken.