What happened
Substack rolled out Reply Rules, reported by TechCrunch, a new moderation tool available to all English-language publications on the platform. Writers set guidelines for how their audience can respond to posts, Notes, and Chat messages, and the system uses machine learning to automatically filter comments that don't match. When a writer hides a reply, the system learns from that action and filters similar content going forward, and writers can review and unhide anything caught by mistake. Substack said the goal is to let creators eliminate "AI slop or profanity, or something silly like only requiring replies in haiku form," positioning the feature as part of its decentralized, writer-managed approach to moderation.
Why Reply Rules matters for founders using Substack
MagnetizeX builds founder visibility systems for B2B firms.
Founder-led newsletters live or die on whether the comment section feels worth reading, and most founders currently either ignore moderation entirely or manually delete a handful of bad replies after the fact, well after the damage to the thread is done. Reply Rules turns that into an upfront setting instead of ongoing cleanup, which matters most for founders who've built a large enough list that AI-generated spam comments and low-effort replies are starting to crowd out real reader engagement. A newsletter with a genuinely good comment section is also a distribution asset in its own right, since readers who see thoughtful replies from other subscribers are more likely to subscribe themselves and share the issue onward.
- Set explicit reply guidelines this week, even simple onesReply Rules only works if writers define what they want, and most haven't set anything yet. Write one clear rule for your Substack this week, such as no links or no one-word replies, so the feature has something to enforce before your comment section attracts more automated noise.
- Audit your last 90 days of replies for AI-generated spamComment spam has become common enough that Substack built a feature specifically to filter AI slop. Scroll back through your recent posts this week and hide any generic, copy-pasted-sounding replies so the system has real examples to learn from before you rely on it fully.
- Highlight one great reader reply per newsletter issueA well-moderated comment section becomes a growth channel when good replies are visible instead of buried under noise. Pick your single best reader comment from the past week and quote it at the top of your next issue, crediting the reader by name, to signal the kind of engagement you actually want more of.
By the numbers: Reply Rules launched available to all English-language publications on Substack at once, rather than as a limited beta, according to the company's announcement.
What to do this week
Open your Substack settings and turn on Reply Rules with one specific guideline that reflects the tone you want for your comment section, whether that's banning links, requiring on-topic replies, or filtering low-effort one-liners entirely. Then spend ten minutes reviewing your last three posts' comments and manually hiding anything that violates the rule, so the model has real training signal from your own community before your next issue goes out to your full subscriber list.