For me, it's no different than if I DoW them, I would expect them to gather their military forces and pitch them against me instead of just having them sit by passively while I take their cities.
They already do it. They just put a lower priority on this than the human does, which is necessary for balance (no matter how good your economy, if 12-14 AI could prioritize bribing the CS the human would rarely get or keep any). They get more CS from some UAs or tenets/SP, or by fulfilling quests (the passive ones) than they bribe for them. They also organize coups and rig elections. But I see bribes often enough (200 or 500, less often 1000), especially if I took a CS from them, though their favorite answer to that when they have one available seems to be to park a spy. But it happens often enough that an AI will buy back its ally that there must be a code somewhere that, under some conditions (how important the benefits were to their victory conditions etc.), triggers them to pile up the gold to bribe and get it back. They don't do it systematically and even less often they'll do it repeatedly when they didn't reply by parking a spy. This also happens much more often in the few turns before sessions, but the AI seems programmed to make its moves to increase or keep its # of delegates around the 5-turn announcement mark rather than on the last 2 turns, leaving the human a chance to buy them back if he planned accordingly.
It's a fine balance to keep, and it's pretty balanced now You just have too look at how frustrated some players get over Alex's UA (to the point for some of giving up playing with the CS when he's in the game. A lot of people also disable the Diplo Win, not intending to pursue it but not wanting to have to block Alex) to get an idea of how annoyed people would get if they increased the likelihood of the AI using bribes to keep CS allies, or made the AI's decisions about CS too "strategic".
To me it's clear that the solution isn't in this direction. It would become too much a gamble on the higher levels: if you stood a big chance to lose your delegates on the last turn if there remains enough AI with gold, it would happen too often that you can't win by the second attempt for WL, and by that time the AI pursuing CV or SV would have won.
The problem pre-patch was that it didn't happen often enough that you couldn't win at the first attempt (which should be difficult or require active planning) because the @ of delegates was too low, and just having most or all the CS as allies was usually enough, and players often ally the CS for their other benefits. Thus diplo victory happened too early and almost by default, no matter your play style.
They had tried to introduce enough ways to acquire delegates and to spread them enough over the game to make the pursuit of the diplo win follow more the pattern of the CV and SV, but like the International Space Station etc. those were too late, unless they re balanced things to delay the Diplo Win. The patch hasmade it unlikely that the diplo victory happens on the first attempt without some active work from the human (getting the FP, passing World Ideology, resurrecting dead Civs or liberating CS), which is the main thing they needed to solve about it. Nerfing Consulates was the other thing they had to do, as it made it more likely the human wouldn't virtually
have all the CS as friends, and most as allies.
Now it's at the second attempt for WL it's "easy" to win by diplo , after researching Globalization (but it's like saying it's easy to win by Culture if you have the Internet... it's about getting there). That's around the time you might need to fall back on winning by diplomacy if you haven't manage to outrace the AI to CV/SV and it's either that or smash the AI to stop it.
And if there's an unstoppable culture or science runaway, you can then work hard to try to win on the first attempt, or again you plan to destroy the runaway.
So it's not broken anymore. It's still not very enthralling as it's still not a proper diplomatic victory and doesn't satisfy much the players who like a more diplomatic game (like me), but that's a separate issue. Balance-wise it's almost there now, the patch solved most of the issues. It happens more around the same time you can win by CV or SV, and it still work as a "fail-safe" victory, an alternative to destroying the AI leader to stop it from winning. The AI got the Great Firewall, or will beat you to Nanotech and the last spaceship part? Give up, go Globalization, pile the cash, plan specific wars to resurrect a dead civ or two. It's the peaceful Plan B when you can't win and must block a victory.
