Updated December 16, 2020: Microsoft has updated the rollout timeline for this release and add additional details for clarity.Microsoft is making some changes to how tenant (Anti-spam/Hosted Content Filter policy) and user (Safe sender) allows work when it comes to high confidence phish. A message is marked with the high confidence phish verdict when detonated and know that it is malicious. Microsoft wants to ensure that customers are protected and therefore block those messages from getting to the inboxes of end-users. This is normally the case, but tenant and user overrides can stop this from happening. Microsoft has decided to no longer honor Allowed senders or domains when the messages are considered as high confidence phish.
Note: Secure by default will not impact your antispam policy high confidence phish action settings but actually enforce it by ignoring the above stated overrides when it is high confidence phish. In other words, high confidence phish will still go to where you configure it to go, whether it is sending it to quarantine (recommended) or to the Junk Email folder.
Timing: Beginning mid-December through the end of February (previously January)
Action: Review and assess impact
How this will affect your organization:
When this change is implemented Microsoft is going to update filtering rules so that inbound messages that are considered high confidence phish and destined for Office 365 mailboxes will not honor anti-spam policy or Safe sender allows. Emails with other verdicts like (regular) phish, spam, etc will not be affected and the allows will still work as expected.
Note: adding senders and domains to an allow list is not best practice and should be considered as a legacy way of filtering.
What you can do to prepare:
Administrators should use the submission portal to report messages whenever they believe a message has the wrong verdict so that the filter can improve organically.