I've no idea how that leak test works, so drawing serious conclusions seems impossible to me. It's totally possible that some Tor exit node being used by the Tails system under test themselves use Google DNS servers, which would explain the observed results (that I can confirm, by the way).

To comment 3:

dnsleaktest.com is a very well reputed site and recommended by VPN's for DNS leak tests.

This is great, but it does not make it change the way Tor works, that is: you depend on Exit nodes to do DNS resolution for you.

I've run this test using VPN and there were no Google hosts.

This tends to indicate your VPN provider does not use Google DNS servers as their upstream.

It is not possible that all exit nodes use google DNS.

Correct. The results you've mentioned (and that I've reproduced) only tend to indicate that some exit nodes use Google DNS as their upstream, which I find not surprising at all. I don't see only Google DNS servers IP in that dnsleaktest.com.

The same results come up using Tails no matter what exit node is used.

I cannot reproduce this. Care to explain how exactly one may reproduce this experiment?

That leads me to conclude that it is a Tails issue, and a very serious one.

I doubt it, until some reproducible test procedure is described. Bonus points if it does not rely on what some blackbox service tells you.

To comment 4:

Can someone explain why it is dangerous to use Google's DNS?

I don't think it's dangerous to have a portion of Tor exit nodes use Google DNS servers.