But hey, Google has managed to make its spellchecker unlearn the difference between "its" and "it's", I suspect by plugging it into a language model. I'm not here to say it's all good.

Reminds me of 4chan figuring out how to mess with how Google was transcribing books when Google was effectively crowdsourcing transcription through recaptcha at the same time m00t had to add it to the site to combat spam bots.

Recaptcha would show two words, and with some practice it was pretty easy to tell which one was the one it already knew and which one was the "new" word it was looking for human help to transcribe. They found that it would accept any input for the "new" word. So 4chan users started using the n-slur, because of course they would, and they actually had enough people doing it to effect some books Google was transcribing.