It would probably make no sense from the lore perspective, but it would be great if the wildlands get bigger in act II and III to threaten all players. They would work like the random encounter in MOO that attacks the players later in the game.

I was actually thinking exactly that when playing Beta 1 yesterday - I really wanted the wilds to be a bit more aggressive.  To be honest, I felt like the wilds stuff was just waiting there for me to farm it for most of the game.  I think the complete lack of critters intruding into your domain may be a bit overly safe.  I remember noticing at some point (slightly alarmed) that there was a ginormous drake army around five squares from my capital.  I then realized that it was doing nothing, and I ignored it for a while until it was convenient to bring my Hero of Doom back to deal with it.  I think it would make a lot of sense in middle to late game for some monster armies to spawn that will actually attack your cities - it would make it a lot harder to run full econ and not bother defending cities.

Two potential things that could help with that:

1. We need more/better defender advantage.  I know that there are a lot of mechanisms already intended to create this, but in my Beta 1 games it didn't feel like any of them were particularly effective (in the hands of the AI).  The effect of this is to make on-the-fly counterattacks with your stack of doom the easiest way to respond to aggression, and means that you don't really need to prepare much to take a city.  Part of that problem may just be the mismatch between champion strength and regular army strength - this may have changed in Beta 2, it sounds like it may be harder to get champions up to the point where they are a stack of doom entirely on their own, and that's largely where I found the defender's advantage breaking down (turn 1, walk up to their archers, kill one stack.  turn 2, kill another stack of archers.  Watch as they flail ineffectually without hitting your champion with >100 dodge, and then get killed by the counterattack.  turn 3, kill another stack...)  That's a very delicate balance, since a high level champion or sovereign really ought to feel obscenely overpowered compared to conventional units.  

As a suggestion, one thing that I liked about Master of Magic was that when your town had walls, it actually translated into a tactical advantage - the walls actually existed on the battlefield and obstructed movement and you could defend a bottleneck while your archers stayed safe and unflankable behind walls. (Well, unless they could fly.  Or teleport.  Or break down walls)  While the defensive bonus that walls give currently is nice, it's completely in the background and doesn't alter tactical combat in the way that having city walls ought to.  In the absence of actually adding a specific wall mechanic to tactical gameplay (although it would be kinda cool to have a wall-climbing ability for champions in the assassin path...), walls could perhaps just create impassable terrain on the battlefield in such a way as to create a one-square bottleneck that the attackers would need to make their way through.  If militias were also given some form of ranged attack (maybe something like thrown rocks, upgraded to bows when the tech is discovered) that would make it significantly harder to conquer a city without having to do prep work.  If the AI were also taught to put a unit with high def/dodge in the bottleneck and use the full defensive command, it might work even better.  (there are all sorts of more complicated ways in which something like walls could be implemented, and many of those would be better - this is an example of doing it without hanging the game engine)

2. The rankings need to better reflect champion strength (including itemization).  In the game I played it was the rating that was wrong rather than the opponent's strength- my sovereign and one other champion had cleared out all the monsters on a very large chunk of the map while I stayed on only one or two cities, and had reached the point where they were both very high level, very well outfitted, and laughed in the face of the far inferior troops my opponent had.  However, since this wasn't reflected in the empire power rating (and therefore presumably wasn't reflected in the diplomatic AI's calculations) I got declared on and the opponent ended up entering into a war he hadn't a prayer of winning.  Champion strength should probably have an exponential effect on empire power rating.  It's like having nuclear weapons in the real world - it doesn't matter how large a country is geographically, if they have nuclear weapons they get treated as a serious player.  The in game empire power ratings need to have a similar understanding of the game world - the largest army and empire are good, but if the other empire is packing the FE equivalent of nuclear weapons, the big empire should be thinking twice about declaring on the small one.  Whatever the actual power curve and difficulty of leveling champions ultimately is, if they reach the point of becoming veritable demigods, the system should do its best to recognize that and reflect it in the perceived strength of the empire.

For that matter, two other things that need to be given more weighting in terms of perceived empire power are money and key defender's advantage technologies.  Cash should almost be treated as if it was equivalent to several turns of rush buying the best defenders I have available, and certain defensive techs (most importantly archery) are going to make the defender much better.  Enemy aggression is good, but their opportunism needs to be calibrated so that they're only opportunistic when they can actually win the battles.