Substack adds an AI detector to help spot blogs written by no one

This reads less like a new revenue stream and more like a defensive trust layer. The strategic signal is that platforms monetizing human voice are starting to treat AI contamination as a retention and brand-safety problem, which should modestly improve pricing power for communities where authenticity is part of the product. The incremental spend is unlikely to move near-term financials, but it can matter for churn, paid conversion, and advertiser comfort over 6-18 months.

Second-order, the biggest winners are platforms with strong identity or closed-network moats; the losers are open comment/feed ecosystems where moderation costs scale faster than engagement. That argues for higher operating leverage at smaller community platforms relative to broad social networks that must police much larger volumes of synthetic text. The key risk is that detection is noisy: false positives can alienate legitimate creators, so a poorly calibrated rollout could hurt participation before it helps trust.

The contrarian miss is that AI detection itself is probably not a durable profit pool. It is likely to commoditize quickly unless paired with provenance, identity, or payment rails. So the market should not extrapolate this into a meaningful standalone AI/software TAM; the real economic value is in who can prove human quality at scale, not who can label text most accurately.