Spam detection on newsletters too strict

problem
1

It seems newsblur has at some point implemented various anti-spam checks on incoming newsletters, which are now causing false positives and rejecting legit newsletters. I run my own mailserver, so incoming newsletters go there and are then bounced to newsblur. Forcing things to ipv4 and making sure SPF is set up mostly works to get my newsletters past spamhaus and improvmx, but I’m still getting failures on some.

Given the access key required in the newsletter address, I’m not sure any spam detection should really apply if the recipient is valid.

Sample rejection message on a legit newsletter:

SMTP error from remote mail server after end of data:
550 5.7.1 Message considered as SPAM (Score of 5.2/5 with BAYES_00,

DEAR_FRIEND, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DMARC_PASS,
HELO_DYNAMIC_IPADDR2, HTML_MESSAGE, MIME_QP_LONG_LINE, RDNS_DYNAMIC,
SPF_HELO_NONE, TVD_RCVD_IP, T_KAM_HTML_FONT_INVALID) - ImprovMX v2026.01.28

2

I’ll add that the most recent bounces were originally sent from Bandcamp, Kickstarter, and Substack. Most newsletters from them are going through, so it seems the actual content must be tripping the Bayes spam score too high.