I have only driven AC on my local machine, not ACC which is a different product so take any comparison with a grain of salt.
iRacing has some fairly large discrepancies with actual physics. Based on what you will read online, most iRacing folks will just learn to live with the problems and move on because the series format and online organization components of the game are there (notice I did not say they are "good" because they are run like a local YMCA youth league at best). That being said, for online series racing, they are the most popular, I'm just saying that iRacing is not a great simulation of an actual car.
From a physics standpoint iRacing is not great. The on-track physics are pretty livable but FFB seems too light for many scenarios and I end up having to make sacrifices all over the place if I try and fix those. No big deal, it's a video game but ironically I find AC to be the opposite in that many scenarios, the FFB is too heavy and fixing them has the similar but different problems from iRacing. Off-track physics are not correct in iRacing at all, so if a wheel goes off track the wheel will completely loose traction, it feels like they are only considering the coefficient of friction of the top of a hard surface without considering things like the "softness" of the outside of the track etc. In iRacing, you can drop a wheel off on a straight and immediately go into a completely uncontrollable spin which would typically require some relatively aggressive input on the driver's part in real life. It feels like they are purposefully trying to penalize you for going off track but based on a blog post from one of their engineers a few years ago, it sounds like they just don't bother with any detailed analysis of how cars behave once they leave the track and ONLY change the coefficient of friction with no other analysis which would probably explain the poor off-track physics.
Another example of the wonky physics is the crashing. You will periodically see cars flying through space in trajectories that do not match either the speed or direction of the original crash. I witnessed this just this past Saturday on the Jefferson circuit on Summit Point. This was a baseline Mazda event with fixed setup (not that it would matter), the stock iRacing Mazda MX-5 on this track rarely breaks 100 and when it does it only does so for a handful of seconds at best. Lap 1 of the race someone in the "pack" managed to crash into another car on-track and somehow flew through the air and hit a car that started from the pits and was two turns back. It was absolutely amazing to watch but very likely not physically possible without some type of propulsion system to generate the height and distance achieved.
Overall, I like driving cars on AC more than iRacing because they feel like the real car and behave as expected for the most part (for the cars I have driven on track). For wheel to wheel racing, iRacing wins for me because of the ecosystem, not the car, not the car dynamics, not the graphics, not the reality or any other technical measure other than being a more mature racing series with a backend designed around it.
All that said, ACC could be a completely different animal, it would seem silly to me that they would use a different physics model across the two games but I guess it's possible.