AGC reacts to an AM signal a bit differently from how it reacts to an SSB signal, depending on filter settings. I've noticed this over the years going back to PowerSDR/OpenSDR-MRX. It's not a bug, is probably working as designed, but could be better. The symptom is that the audio level of an AM signal is sometimes noticeably lower in AM or SAM mode than it is when using an SSB mode at zero-beat for the same signal.
To see this, try the following on 80m where good AM signals are plentiful: Set your LSB filter's upper bound to something below the carrier, like -50 Hz (the lower bound can be something reasonable for LSB, like -3000), and set your SAM filter width to something reasonable for AM, say 6 kHz (-3000 and +3000) or so, whatever it takes to exclude QRM. Select LSB mode and tune in an AM signal to zero-beat - i.e. the VFO is right on the signal's carrier frequency, and observe the audio level by ear. Now, without changing anything else, switch to SAM mode. You should notice that the audio level is lower. You can switch back and forth between SAM and LSB and hear the difference.
I believe what is happening is that in SAM mode the AGC is reacting mostly to the carrier, which is the highest level component of the signal, whereas in SSB it's reacting only to the SSB envelope, which is lower than the carrier on average (just look at the panadapter). Again, this is not a bug but is the way it is expected to work.
You can force the two audio levels to be the same by doing the following: Change your LSB filter to have an upper bound that includes the carrier, say +50 Hz. Now tune in the same AM signal on LSB to zero-beat as above and switch back and forth between LSB and SAM. You should observe that the audio level is now equal between the two. That's because in LSB the passband now includes the carrier and the AGC is now reacting to the carrier (even though it's at zero beat and you can't hear it) just like it is in SAM mode.
This is not a big deal but it is a minor annoyance that switching between modes requires me to adjust the AF gain. I wonder if the AGC can be made to react to the sidebands alone in AM modes to mitigate this effect. And perhaps make that a user-selectable AGC setting. I haven't looked at the code myself (yet) to see how it works, but I suspect anything's possible with SDR.
If this were implemented, I'm guessing that you might lose the "quieting" effect of the carrier when in AM mode, but SAM would work as before (or not?). So perhaps a compromise would be to have the AGC gain be different on AM and SSB for a given signal based on whether or not there is a carrier.
Just tossing this out there for discussion.