As of DXP, a new validation process was added to check if links in a web content article links to a valid Liferay page. The way the validator works is this: Once it detects a link within the web content article's content, it first checks the beginning of the link to determine what the domain is. If the domain matches one of the virtual hosts configured by Liferay, it will strip the domain (and port, if it exists) out of the URL and send the tail end of the URL to the validator. Otherwise, it assumes that the link is going to an external page, rather than to a Liferay page, so it skips validation for that link.
The problem is, someone could have their server configured to consist partially of Liferay pages and partially of external pages, within the same domain. So we cannot assume that just because the domain matched up with one of Liferay's virtual host names, that it is supposed to be a link to a Liferay page.
The only solution that I can think of for this problem is to add a configuration option to skip this validation process. If a user has their server set up in this way, then they should be able to disable this validation process so that they can publish and import content that has links to pages on the same domain name but outside of Liferay.