YouTube, like most social media platforms, uses Artificial Intelligence (AI) to detect spam subs. Questions like this are impossible to answer because a key characteristic of AI is that it learns constantly and changes the underlying model based on new data. So if all of a sudden with in a 2-day period the AI picks up a bunch of subscribers from Asia that watch videos for exactly 32 seconds, it may decide that is SPAM and start removing subs that meet that criteria.
The "algorithm" the AI builds is YouTube/Google's intellectual property and is very valuable. So they do not ever provide details of how it works publicly.
Anything anyone tells you about how the anti-spam system works is a guess based on trial and error and may not be valid for more than brief period of time.