The important part of his statement to me was:

>on my site

That means not uploaded to github. That means self hosted, as is the point of the main discussion.

>these scanners don't respect the LICENSE file

I don't think github scans outside repos, but what is stated there certainly applies to OpenAI and others. They don't have a license to do what they are doing, but the US is not enforcing copyright law on them out of fear of losing the AI race.