I've been hassling ZWO about this for a while now, had posts removed and am currently on a 30 day timeout from one of their Facebook groups for bringing it up again. I've done forensic teardowns of the ASIair Pro, as well as ripped apart the update packages. I've emailed ZWO, I've DM'd the owner on Facebook, nothing. Silence.

Here's what I've found. The ASIAir 1.4.0 image you can download from their website is packing indilserver 1.8.2. Doing an update from the ASIAir 2.0 app, using the current 10.46 build, does some fun stuff. They're using APT for package management, and the update grabs a bunch of updates from a Chinese raspberrypi.org mirror, except for five packages, which I believe are transferred to the ASIair directly from the tablet application:

Get:16 /tmp/zwo/update/deb/asiair_armhf.deb asiair armhf 1.0.0-1046 [9,138 kB]
Get:17 /tmp/zwo/update/deb/indi_stargo-y_armhf.deb indi-stargo armhf 1.5-2 [138 kB]
Get:18 /tmp/zwo/update/deb/indi_starbook-y_armhf.deb indi-starbook armhf 0.8-1 [681 kB]
Get:19 /tmp/zwo/update/deb/indi_eqmod-y_armhf.deb indi-eqmod armhf 0.7-6 [448 kB]
Get:20 /tmp/zwo/update/deb/indi-y_armhf.deb libindi1-y armhf 1.7.8-22 [8,974 kB]

The first, the asiair_armhf_deb package, is the core runtime apps that drive the guider, imager, and updater, as well as the shell scripts for managing the wifi and other essentials. The -1046 minor version tag matches the 10.46 version of the tablet app.

If you're using the INDI kit, he other four will look pretty familiar, except for the -y suffix.

Pulling the update packages off to another host because the asiair is usually running a read-only file system and the tools I need for this aren't installed there:
billn@Hadriel:~/asiair-dpkg$ scp pi@*-y*.deb .
indi_eqmod-y_armhf.deb                                                                                                       100%  438KB  34.9MB/s   00:00
indi_starbook-y_armhf.deb                                                                                                    100%  665KB  38.1MB/s   00:00
indi_stargo-y_armhf.deb                                                                                                      100%  135KB  32.5MB/s   00:00
indi-y_armhf.deb                                                                                                             100% 8764KB  38.6MB/s   00:00
billn@Hadriel:~/asiair-dpkg$ scp pi@*.deb .
asiair_armhf.deb                                                                                                             100% 8924KB  35.3MB/s   00:00
billn@Hadriel:~/asiair-dpkg$ for i in `ls`
> do
> mkdir $i.d
> cd $i.d; ar -x ../$i; cd ..; done
billn@Hadriel:~/asiair-dpkg$ ls
asiair_armhf.deb    indi-y_armhf.deb    indi_eqmod-y_armhf.deb    indi_starbook-y_armhf.deb    indi_stargo-y_armhf.deb
asiair_armhf.deb.d  indi-y_armhf.deb.d  indi_eqmod-y_armhf.deb.d  indi_starbook-y_armhf.deb.d  indi_stargo-y_armhf.deb.d

So that's all five packages unpacked. Cracking open indi-y:
billn@Hadriel:~/asiair-dpkg$ cd indi-y_armhf.deb.d/
billn@Hadriel:~/asiair-dpkg/indi-y_armhf.deb.d$ ls
control.tar.xz  data.tar.xz  debian-binary
billn@Hadriel:~/asiair-dpkg/indi-y_armhf.deb.d$ tar xvf control.tar.xz
billn@Hadriel:~/asiair-dpkg/indi-y_armhf.deb.d$ tar xvf data.tar.xz

I can't readily find a libindi1-1.7.8 debian package to compare this to, but at the most basic, we have this:
billn@Hadriel:~/asiair-dpkg/indi-y_armhf.deb.d$ cat control
Package: libindi1-y
Priority: extra
Section: checkinstall
Installed-Size: 17268
Maintainer: root@raspberrypi
Architecture: armhf
Version: 1.7.8-22
Provides: libindi
Description: Instrument-Neutral Device Interface library -- shared library
Replaces: libindi-plugins, libindi0, libindialignmentdriver1, libindidriver1, indi-bin, libindi-data, libindi1
Conflicts: libindi-plugins, libindi0, libindialignmentdriver1, libindidriver1, indi-bin, libindi-data, libindi1
Depends: libc6,libgsl23,libgslcblas0,libstdc++6,libgcc1,libusb-1.0-0,libnova-0.16-0,libcfitsio7,zlib1g,libjpeg62-turbo,libudev1,libcurl3-gnutls,libbz2-1.0,libnghttp2-14,libidn2-0,librtmp1,libssh2-1,libpsl5,libnettle6,libgnutls30,libgssapi-krb5-2,libkrb5-3,libk5crypto3,libcom-err2,libldap-2.4-2,libunistring2,libhogweed4,libgmp10,libgcrypt20,libp11-kit0,libtasn1-6,libkrb5support0,libkeyutils1,libsasl2-2,libgpg-error0,libffi6,libcurl4,libssl1.1

22 revisions on 1.7.8, if I'm reading that correctly. I've done this teardown before, but I can't find my dumps of it to compare against an older version. I'll keep digging around, I'm sure it's here somewhere. If someone has an ASIair that hasn't been powered on in a while and wants to dropbox me an image of the SD card, I can tear that open for the last update and compare.

The last time I was digging around in here, I was trying to sideload drivers for my Sesto Senso 2 and Esatto EAFs, figuring it's an indiserver, I expect a certain level of interoperability, maybe I can trick it into using my focusers, or at least use a separate app to control them and get the functionality I want without buying a ZWO EAF when I already have two excellent ones. However, this didn't work, and I couldn't figure out why until I started stack tracing the indiserver and drivers to sort it out. Restarting the indiserver with my specifically requested drivers loaded, calls to the focuser weren't going to the USB serial device for either focuser, but were instead forcibly redirected to the tcp port of the zwo_imager application.
root@asiair:/home/pi# netstat -plan | grep LIST
tcp        0      0  *               LISTEN      732/zwoair_imager
tcp        0      0   *               LISTEN      1069/smbd
tcp        0      0  *               LISTEN      718/zwoair_guider
tcp        0      0  *               LISTEN      568/zwoair_updater
tcp        0      0  *               LISTEN      732/zwoair_imager
tcp        0      0  *               LISTEN      732/zwoair_imager
tcp        0      0  *               LISTEN      718/zwoair_guider
tcp        0      0  *               LISTEN      568/zwoair_updater
tcp        0      0   *               LISTEN      1069/smbd
tcp        0      0  *               LISTEN      718/zwoair_guider
tcp        0      0  *               LISTEN      718/zwoair_guider
tcp        0      0    *               LISTEN      384/sshd
tcp        0      0  *               LISTEN      653/pigpiod

I haven't done too deep a dive into what's going on for each of the different listening socks for each app but both the guider and imager apps have some familiar bits inside of them:
billn@Hadriel:~/asiair-dpkg/asiair_armhf.deb.d/home/pi/ASIAIR/bin$ strings zwoair_imager | grep -i phd
billn@Hadriel:~/asiair-dpkg/asiair_armhf.deb.d/home/pi/ASIAIR/bin$ strings zwoair_guider | grep -i phd
Your %s is preventing PHD from making adequate corrections to keep the guide star locked. Increase the %s to allow PHD2 to make the needed corrections.
Even using the maximum moves, PHD2 can't properly correct for the large guide star movements in %s. Guiding will be impaired until you can eliminate the source of these problems.
Calibration step = phdlab_placeholder, Max RA duration = %d, Max DEC duration = %d, DEC guide mode = %s
PHD2 version, Log version
PhdController failed: %s
PhdController: newstate STATE_FINISH
PhdController: setup
PhdController: newstate STATE_ATTEMPT_START
PhdController: not ready: %s
PhdController: guider state = %d
PhdController: newstate STATE_SETTLE_BEGIN
PhdController: newstate STATE_CALIBRATION_WAIT
PhdController: start capturing
PhdController: newstate STATE_SELECT_STAR
PhdController: newstate STATE_CALIBRATE
PhdController: newstate STATE_WAIT_SELECTED
PhdController: clearing calibration
PhdController: start calibration
PhdController: newstate STATE_GUIDE
PhdController: setting Dec guide mode to %s for dither settle
PhdController: newstate STATE_SETTLE_WAIT
PhdController: settling, locked = %d, distance = %.2f (%.2f) aobump = %d frame = %d / %d
PhdController: restore Dec guide mode to %s after dither
PhdController complete: success
PhdController complete: fail: %s
PhdController: newstate STATE_IDLE
PhdController::Guide reentrancy state = %d op = %d
PhdController::Guide begins
PhdController: newstate STATE_SETUP
PhdController::Dither reentrancy state = %d op = %d
PhdController::Dither begins
PhdController: forcing dither RA-only since Dec guide mode is %s
PhdController::Dither pMyDev->Dither failed
manual or phd1-style dither

So it looks like the guider started out as Phd2, which is distributed under the BSD license. Looking at the guider log, we see a few familiar functions from Phd2, with extensions that make sense to support the tablet app.
 1-23_23:11:02.116673 [handle_cli_input_event]->4400:{  "id" : 56,  "method" : "get_ra_dec_history"}
 1-23_23:11:02.116816 [do_notify1]<-4400(0):{"jsonrpc":"2.0","method":"get_ra_dec_history","error":"history is empty now","code":321,"id":56}
 1-23_23:11:02.117046 [Wait]join thread success
 1-23_23:11:02.117154 [Wait]join thread>
 1-23_23:11:02.118127 [handle_cli_input_event]->4400:{  "id" : 57,  "method" : "get_camera_binning"}
 1-23_23:11:02.118321 [do_notify1]<-4400(0):{"jsonrpc":"2.0","method":"get_camera_binning","error":"device not connected","code":318,"id":57}
 1-23_23:11:02.118515 [Wait]join thread success
 1-23_23:11:02.118624 [Wait]join thread>
 1-23_23:11:02.119311 [handle_cli_input_event]->4400:{  "id" : 58,  "method" : "scope_get_track_state"}
 1-23_23:11:02.119458 [do_notify1]<-4400(0):{"jsonrpc":"2.0","method":"scope_get_track_state","error":"mount is not connected","code":315,"id":58}
 1-23_23:11:02.119651 [Wait]join thread success
 1-23_23:11:02.119760 [Wait]join thread>
 1-23_23:11:02.121107 [handle_cli_input_event]->4400:{  "id" : 59,  "method" : "scope_get_ra_dec"}
 1-23_23:11:02.121352 [do_notify1]<-4400(0):{"jsonrpc":"2.0","method":"scope_get_ra_dec","error":"fail to operate","code":207,"id":59}
 1-23_23:11:02.121588 [Wait]join thread success
 1-23_23:11:02.121733 [Wait]join thread>
 1-23_23:11:02.121813 [handle_cli_input_event]->4400:{  "id" : 60,  "method" : "scope_is_moving"}
 1-23_23:11:02.121950 [do_notify1]<-4400(0):{"jsonrpc":"2.0","method":"scope_is_moving","error":"mount is not connected","code":315,"id":60}
 1-23_23:11:02.122144 [Wait]join thread success
 1-23_23:11:02.122299 [handle_cli_input_event]->4400:{  "id" : 61,  "method" : "get_flip_calibration"}

Digging through strings output of the 1.7.8 indiserver in /usr/bin nets us some gems as well:
pi@asiair:/usr/bin $ strings indiserver | grep indi | more
root@asiair:/usr/bin# strings zwo_deleteStarsTool  | grep Siril
FITS image format %d is not supported by Siril.

So, yeah. Binary distributions of modified source trees for more than one open source kit. I don't know enough about licensing for the third party drivers, but it appears a few of those have been modified as well, from reading other forums where folks have had issues with those particular mounts. If they're contributed to the indi-3rdpart repo, that's also LGPL-2.1, so source/changes needs to be provided for those as well. Siril is GPL3. Phd2 is BSD-3, which requires disclosure of the license and copyright notice.

If an INDI server is running on something, I expect a certain behavior profile, like being able to load drivers and have them work. As configured, the indiserver distributed by ZWO on the ASIAir is deliberately configured to prohibit the use of hardware from competing vendors, which is anti-thetical to what I consider the open source spirit to be. Another word for that is 'anti-competitive'.

This is extra shady because ZWO has INDI drivers for their hardware, but they don't use them on the ASIair. They aren't even present on the unit. If ZWO wanted to leverage the indiserver legally in their product, they could ship it with everyone else's drivers removed, eat their own dog food, and make sure their INDI drivers are best of breed. Just my two cents.