>>4175 (OP)
I'll repost my prior suggested solutions:
1) Auto-spoiler content if it receives a sufficient amount of CP reports.
2) Allow mods to manually disable reports for posts and content (based on hashes) which are deemed to be false positives, in the rare event of abuse. Even with malicious actors trying to exploit this system, this takes a lot of effort to do with little pay off for them especially with the second feature.
3) Continue to data-mine true positive hashes, for training the pre-filter AI, without saving any actual content for obvious legal reasons.
4) Data-mine past and future true positives to detect patterns related to the text content, which will give strong hints if a post will be uploaded with bad content as a pre-filter.
5) Instead of disabling all links as a whitelist, employ a banned blacklist if there's a pattern discovered with what sites are linked.
6) Consider a warning to anons before re-directing them to links. This will seem obtrusive, but it is my suspicion that the offenders are hoping to catch IP addresses which accidentally click with their cursor when either scrolling on the Overboard or when filing a report. This would prevent any such incidents, for the most part, directly foiling our assailants' stratagem.