Automation rule triggering even if criteria doesn't match??

Hi, I have automation rules to handle preference page options. It checks/unchecks the checkbox fields in pardot that syncs with SF fields.

Scenario 1 : The prospect ends in a preference list when prospect selects option in preference page. AR1 then checks the checkbox field in pardot that syncs to Salesforce. (Repeat Rule checked)

Scenario 2: BU unchecks the checkbox in salesforce that syncs with pardot. AR2 then removes the prospect from the preference list. (Repeat Rule checked)

Issue: When AR2 is executed, AR1 also executes which is weird. that results in data mismatch.

Language on the preference page is opt out model

- I no longer want to receive newsletter.

Below are my ARs:

AR1: Newsletter - Opt Out via EPC

Rules:  Prospect public list opt-in status | Prospect is not opted out of the list 01. Newsletter List for EPC

Actions:  Change prospect custom field value | Opt out of Newsletter to 'true' 

AR2: Newsletter - Opt In via SF

Rules: Prospect custom field | Opt out of Newsletter is 'false'

and Prospect list | Prospect is on list 01. Newsletter List for EPC

Actions: Remove prospect from list | 01. Newsletter List for EPC

Step to replicate:

Prospect opt outs, AR1 sets the checkbox true, syncs the checkbox to SF.

BU then opts back the contact record in SF to false (opt-in), AR2 removes the prospect from the list, at the same time AR1 also executes that sets the checkbox to true and syncs back to salesforce.

This is driving me crazy, how Automation Rule criteria runs. Is there a better way to handle this scenario? I have looked into adding tags in the logic but the back and forth changes from preference page and Salesforce makes it more complicated.

If anyone could pitch in and help would be greatly appreciated. 

Thanks in advance.