Substack’s new tool tells you who’s been writing their newsletters with AI

Substack has rolled out a feature that estimates how much of any given newsletter was drafted with AI. The mechanism is straightforward: readers see a signal, writers know the signal exists, and the platform collects the data either way.

The interesting question is why now. Substack built its reputation on the premise that individual writers, not institutions, are the editorial authority. That premise gets complicated when a significant share of the prose on the platform may have been generated by a model trained on other people's work. A disclosure tool lets Substack preserve the author-first brand while quietly acknowledging that the writing process has changed.

For independent writers, the calculus is simple. A high AI estimate does not necessarily mean low effort; many professionals use AI for research, outlining, or editing. But readers tend to treat the label as a proxy for authenticity. Writers who lean heavily on AI generation will need to decide whether the efficiency gain is worth the credibility discount.

The broader signal is about platform liability. Every content host is now navigating the same problem: how to demonstrate to advertisers, regulators, and users that it has visibility into what is being published. Substack's approach, voluntary disclosure surfaced to readers, is lighter than mandatory labeling but heavier than silence. It sets a baseline other creator-economy platforms will be measured against.

For the labor market, the takeaway is narrow but real. Freelance writing income has already been compressed by AI-assisted competition. A transparency layer does not reverse that pressure; it simply makes the supply more legible to buyers. Writers who can articulate a distinct voice, proprietary research, or a trusted relationship with their audience will find that more valuable. Those selling generic prose at commodity prices will find it less.