Content Moderation Policy

1. Overview

Shivam Das Bulletin News is an open, community-contributed wire — anyone can publish a dispatch. To keep the wire trustworthy for readers and safe for advertisers, every post is automatically checked for spam, hate speech, and violent content as soon as it's created or edited. A post that trips the detector is taken offline right away, rather than left live pending a moderator's schedule, and a human reviews it before it can be republished.

2. What we screen for

The system checks each post against three categories:

3. How detection works

Detection is heuristic, not a black-box AI model — the same explainable, rule-based approach the site already uses for its plagiarism and named-entity checks. Each post's title and body are run against a maintained set of patterns for each category. A match earns a severity score for that category; a post is only flagged once the score for a category crosses a fixed threshold, so a single incidental word match doesn't trip the system. Because the rules are explicit patterns rather than a trained model, every flag can be traced back to exactly what matched — there's no unexplainable verdict for an admin, or an author appealing one, to work from.

4. Auto-unpublish & moderation hold

If a post crosses the threshold in any category:

• It is immediately unpublished — this happens at creation and again on every edit, so a post can't slip through by being changed after the fact
• The post enters moderation hold, a state separate from a normal unpublish, meaning the author's own Publish/Republish control is blocked while the hold is active — an author can't simply click "Republish" to undo a hold themselves
• A record of which category matched, its severity, and the specific matched text is logged for the review queue. Matched snippets are only ever visible to admins reviewing the flag — never shown to readers
• The author is notified that their post was held for review

5. Human review

Flagged posts sit in an admin moderation queue, ordered by severity. A human reviewer looks at each flag and either:

• Approves it — clearing the hold and restoring the author's ability to publish the post, or
• Rejects it — confirming the violation, keeping the post unpublished

Every flag, resolved or not, stays on record as part of the site's moderation audit trail — history is never deleted, only marked resolved.

6. Appeals & false positives

Heuristic detection can misfire — for example, a news report that quotes a threat while covering a violent event, or a post that happens to contain a pattern the system was tuned to catch. If your post was held and you believe it was a false positive, contact us through the Contact Us page with the post's URL so an admin can review it.

7. Limitations

This system is a first line of defense, not a complete one. Pattern matching can miss content that doesn't use the specific language it checks for, and it can occasionally flag content that shouldn't be held. It is not a substitute for reader reports — if you see something on the wire that concerns you and wasn't caught automatically, please tell us through Contact Us.

8. Changes to this policy

We may update this Content Moderation Policy, and the underlying detection rules, from time to time as new patterns of abuse emerge.

9. Contact us

Questions about this policy, or about a specific held post, can be sent through our Contact Us page.