What is Substack?

Matthew Prince illustrates the AI content monetization problem: OpenAI buys one Substack subscription, ingests all content, then redistributes it to every user, collapsing the per-reader payment model that supports publishers.

“It can't be that OpenAI buys a single Substack subscription to each of their of their publications, slurps all the information out, and then publishes it to everyone. And yet, that's kind of exactly what's happening right now.”

Grant Sanderson on LLMs preferring uninsightful writing over genuinely excellent content.

“Maybe the ultimate failure there is that even if they're good at discriminating between a B essay and an A essay, they're not actually good at discriminating between an A essay and a thing you actually want to read, something that would be followable on Substack and insightful. They actually end up preferring uninsightful pieces of writing.”

Substack added anti-AI tools for writers, with CEO Chris Best signaling this will become a broad platform trend.

“I don't know if you saw last week's Substack added a bunch of anti- AI tools. And Chris Best, the CEO, >> this is going to be this is going to be a big trend.”

Substack's economic incentive for banning AI newsletters: mass AI-generated content with free subscriber email distribution would impose enormous infrastructure costs on the platform.

“I can imagine like they're probably like staring down the barrel of a future where like everyone in every profession is like I'm going to start a newsletter. I'm going to have a chatbot write it or a chatbot respond to all of my comments. And all of a sudden that's going to become an enormous expense for them.”

A static, no-context LLM prompt with no personalization is sufficient to produce useful summaries of newsletters and podcast transcripts, suggesting heavy engineering is not always needed for valuable AI summarization workflows.

“This tool literally just takes the full text of each Substack post or podcast transcript dumps it into an LM and says summarize this and like yeah, I've done a little bit of iteration on the exact prompt like, you know, what kind of information I want it to surface and what I don't, but it's completely static, no context at all.”

More than 40% of long-form content on LinkedIn is AI-generated, versus 29% on X and 10% on Substack.

“Over 40% of long- form content on LinkedIn is now AI generated, much more than 29% on X and 10% on Substack.”

Substack's crackdown on AI-generated newsletters, while continuing to host Nazi content, reveals that platform economics rather than ethics drive content moderation priorities.

“I think it is also like a threat to their business in a way that maybe Nazi content is not.”

The Curve (by Steve Newman) is a personalized AI-powered newsletter aggregator using static LLM summarization of Substack posts and podcast transcripts, worth watching as a lean solo-built AI media product.

“This tool literally just takes the full text of each Substack post or podcast transcript dumps it into an LM and says summarize this and like yeah, I've done a little bit of iteration on the exact prompt like, you know, what kind of information I want it to surface and what I don't, but it's completely static, no context at all.”

Kevin Roose on Substack implicitly ranking AI writing as more offensive than Nazi content.

“In this one moment, Substack has told us that on some level they find AI-generated writing more offensive than Nazi.”