X-Age-Rating would only work if the server could be sure of the jurisdictions under which the recipient was bound.
To continue the thought experiment though: another implementation would be to list up to N tags that best describe the content being served. You could base these on various agreed tagging systems such as UN ISIC tagging (6010 Broadcasting Pop Music) or UDC, the successor to the Dewey Decimal System (657 Accountancy, 797 Water Sports etc.) The more popular sites could just grandfather in their own tag zoologies.
A cartoon song about wind surfing:
X-Content-Tags: ISIC:6010 UDC:797 YouTube:KidsTV
It’s then up to the recipient’s device to warn them of incoming illegal-in-your-state content.
There actually was a proposal/standard for this back in the day: https://www.w3.org/PICS/
> X-Age-Rating would only work if the server could be sure of the jurisdictions under which the recipient was bound.
That's no different to the current legislation.
The twitter API used to have a "illegal in France or Germany" field, which was used for known Nazi content.