The first issue Medium would have to deal with in deciding whether to accept AI-generated writing is, can the average person tell the difference? Is there a way for Medium to determine, from the content if an AI program generated the writing? If not, there seems no point to me in making a rule against it.

I do oppose the use of AI-generated articles here on Medium because the richness of Medium lies in the fact that humans write the content. The content here is individual and flavored with the writers' personalities. Permitting AI would cause Medium to lose that.

As you've said, AI can getfacts wrong. Humans too can get facts wrong, but we at least make an effort to ensure we are correct, something that AI cannot do. AI has no conscience; it works on the GIGO principle.

If a piece of writing is AI-generated, the poster should be required to state that.