1) Pushbullet is frequently 'randomly selected' for extra scrutiny (TSA style) because it competes with some offering from Google or a preferred partner.
2) The review algo simply diffs the resubmission with the previous version and if there are changes 'near' any of the keywords from the violation, it gets approved, until the next 'random' scan.
Pushbullet is directly competing against google's "messages" app. They have the exact same use case. Note that Messages does not display such prominent markers that it's uploading stuf to google server yada yada, probably because it is immune to play store verification.
Or just because it looks at SMS. Few non-malicious apps have a reason to look at SMS and it's very high value data for malicious ones, no surprise the AI model misclassifies the one app with a perfectly legitimate use. It should be whitelisted but hand tweaks to algo results are probably taboo at Google.
1) Pushbullet is frequently 'randomly selected' for extra scrutiny (TSA style) because it competes with some offering from Google or a preferred partner.
2) The review algo simply diffs the resubmission with the previous version and if there are changes 'near' any of the keywords from the violation, it gets approved, until the next 'random' scan.