I don't know much about this topic, so interpret these as questions, not as arguments that I have particular confidence in:
* If the earbuds you use to perform the hearing test are the same ones you use with the resulting EQ profile, why does it matter whether the earbuds are flat or not? If you're taking measurements in a transformed coordinate space, and applying those measurements in the same transformed coordinate space, shouldn't that be equivalent to taking measurements in the original coordinate space and applying them in the original coordinate space?
* Why isn't active noise cancellation a good enough form of noise isolation?
* If using an EQ profile that only matches your hearing loss imperfectly can damage your hearing, wouldn't the same thing apply to EQ profiles that aren't intended to correct hearing, like the media player presets "classical music", "rock music", etc.? For that matter, wouldn't it apply to a flat EQ profile? Is there something special about an imperfect hearing correction EQ profile, compared to other EQ profiles?
> If the earbuds you use to perform the hearing test are the same ones you use with the resulting EQ profile
You also have to account for the microphone, which can (and will) have its own bias. Measuring from a flat response source, and producing for a flat response mic/speaker combination will produce the most accurate results.
> active noise cancellation
ANC creates sound artifacts at low volumes. It also provides, in ideal circumstances, around 25Db of noise reduction. You probably want between 40 and 80Db (or more) of noise reduction, depending on the ambient noise level around you. The threshold of hearing hovers around 10Db for healthy ears. Healthy ears can hearing your own heartbeat in an anechoic chamber (as close to 0Db as we can get).
> wouldn't the same thing apply to EQ profiles that aren't intended to correct hearing
If you're correcting for a 40Db loss at one frequency (taken randomly from the surrounding threads), applying it to a frequency where you have only 20Db loss can result in an extra 20Db of sound at that frequency. 20Db is 4x louder by perception, and over 6x higher pressure.
For a point of reference, Apple music's EQ curves are 12Db, 24Db if you set the extremes. So yes, they could damage your hearing, but that will depend more on the volume than the EQ.
It's probably good enough for an temporary situation, but I wouldn't trust it long term. Improper amplification can damage your hearing further.