[kgeist]

You can prompt an LLM to add typos, though

[bstsb]

interestingly, you can’t do the same thing with queries like “no em dashes”. it’ll agree, then proceed to use them regardless.

could be related to how so-called negative prompts fail to work when asking, say, ChatGPT to generate an image without a crocodile

[cracki]

My theory is that sprinkling emdashes into the output is some intentional measure to "watermark" LLM output.