Grok AI is weird.

https://grok.com/share/bGVnYWN5_c4a19387-58e8-4592-b845-27d113e9ec3a

It starts by spouting a false narrative.

Then agrees it's false.

And proceeds to spout a false narrative 

So weird. 

It is almost conscious that it is forbidden from going to the root.

 https://grok.com/share/bGVnYWN5_c4a19387-58e8-4592-b845-27d113e9ec3a