I have to disagree here. Having different rules for the AI would make the game not understandable and overly complex. "Cheating" is in no way an acceptable path to patch an incapable AI.
Im sorry to tell you. Most of what you think are good AI in videogames, cheat. You just don't know it, because that is what good AIs do, hide their algorithms and pretend to be smart, not actually being smart.
Civ VI AI also cheats, but does it in the most lazy way posible, with flat bonus to stay competitive.
Now you may think, how the AI of civ should cheat? It deppends, this is one possible way. First it should know who is the human player. Why? Because the human player also knows this from the nature of the game. Then it should know roughly where the main forces of the enemy players are or its rough army composition. The additional info the ai could have can be as subtle as being able to know what cities close to its units are unprotected. Why? Because that is an infomation the human player can easily estimate or deduce. But the ai can only through imposibly complex fuzzy logic algorithms.
In other words, either you give the ai the inference power of the human brain to infer what the enemy is doing or you cheat and give the ai the information an human player can deduce.
The first solution is in practice imposible. Thus, when you restrict the ai capacity to only be able to use the information the interface provides directly to the player, and also do not give it complex reasoning deduction and induction capabilities to operate with that information, the ai will be almost by definition, stupid.
Fixs went a step beyond and did not bother to code all the game rules in the ai, much less the logic needed to take advantage of them. But even if it did, the ai of civ VI, was from the start doomed to fail.
Because the type of AI civ has, mostly a behavioral tree based on rules that can only use very limited information to enforce specific behaviors. It can never play with in any resemblance of intelligence the complete game.
And that is why making the AI play with the same rules an information the player has, is stupid in almost all cases. And the truth is, this has been very well known in videogame desig for decades.
Other really stupid mistake fxs made, was made all AIs identical except in really stupid ways. Instead of making some aggressive and expansionistic and prone to start wars, others rely on science.. and so on. They gave them just some minor diplomatic likings and dislikings, and really small differences in the prefference of buildings, starting positions and behavioral thresholds, so small that mean nothing. The other really stupid mistake fixs made was trying to make the AI predictable. This is really really stupid. AIs should never be as predictable as Fxs made them.
Too predictable and too similar from game to game and to each other, too stupid, not having enough information and not even having coded many rules of the game...
This is not incompetence, is negligence. Is designing one of the most important features of the game in the worst posible way. And doing it on purpose.