---
title: "ChatGPT's Publication Share Is 2.5× AI Overviews (Nov 9)"
description: "On the ChatGPT API, gpt-4.1-mini, editorial publications are 40.2% of classified citations, 4,039 of 10,039, against 16.3% for AI Overviews."
canonical: "https://jaxonparrott.com/brief/2026-11-09-chatgpt-api-classified-citation-share"
date: "2026-11-14"
last-updated: "2026-11-18"
---

# ChatGPT's Publication Share Is 2.5× AI Overviews (Nov 9)

By [Jaxon Parrott](https://jaxonparrott.com/about), who coined Machine Relations in 2024 · Machine Relations Index, created by [Jaxon Parrott](https://jaxonparrott.com) · Published November 14, 2026 · Updated November 18, 2026 · [Index release of November 9, 2026](https://machinerelations.ai/findings/editorial-share-of-ai-citations#history). The [finding page](https://machinerelations.ai/findings/editorial-share-of-ai-citations) shows the current release. This Brief stays on the release linked here.

![The editorial share ChatGPT gives publications against Google AI Overviews: 2.5×. Index release of November 9, 2026.](https://jaxonparrott.com/brief-cards/2026-11-09-chatgpt-api-classified-citation-share-16x9.png)

On the ChatGPT API, gpt-4.1-mini, editorial publications are 40.2% of classified citations, 4,039 of 10,039. Google AI Overviews sends 16.3%, 488 of 2,995. That is 2.5 times the share, across the Index's 1,255 buyer questions since May 10. On ChatGPT, editorial publications are the largest class of source. On Google AI Overviews they rank third, behind vendors' own sites at 28.0% and community platforms at 23.3%.

## How wide is the share gap?

It is the widest percentage-point gap between any two engines for any class of source in the release. Of the six engines the Index measures, ChatGPT gives editorial publications the largest share and Google AI Overviews the smallest. I compare shares of each engine's classified citations, not the raw counts. 4,039 against 488 looks like a much larger multiple, and it answers a different question: the two engines do not classify the same number of citations, so the counts cannot stand in for the shares.

A placement in a publication is the class ChatGPT cites most. On AI Overviews that same placement sits behind vendor pages and community threads, and both of those classes outrank editorial there. I call this a share gap on purpose. A reader who treats the citation counts as the comparison gets a different story, and this release does not support that story.

## Which publication leads the Index?

[Forbes](https://machinerelations.ai/index/domains/forbes.com) leads the editorial publications. It is cited in 1,264 of 37,851 runs, 3.34%, first of the 1,388 editorial publications in the release, and cited on all six engines. That 3.34% divides by every run in the Index. The 40.2% divides by ChatGPT's classified citations. One number is a share of the whole Index and the other is a share of one engine, so I keep them apart.

## What would I do about it?

I choose the engine before I choose the outlet. When the buyer asks ChatGPT, the next dollar goes to a publication, and the pitch carries one claim a reporter can print whole: the number, the date and the measurement. When the buyer asks Google AI Overviews, I open the question in the [Index](https://machinerelations.ai/index) and read the vendor pages and community threads it already cites before I pitch anyone. Editorial is that engine's third source. The outlet follows the engine.

## How this was measured

Figures come from the [November 9 Index release](https://machinerelations.ai/findings/editorial-share-of-ai-citations#history): 37,851 answer runs on 1,255 buyer questions across six engines, May 10 to November 9, 2026. The Index called ChatGPT on gpt-4.1-mini. An engine's classified citations are the cited domains in its runs whose source class the Index has assigned, each domain counted once per run. Editorial publication is one of those classes, and unclassified sources are left out of every share. Whole-Index rates such as Forbes's divide by every run. A script computed every figure here, and the standing page recomputes the contrast whenever a new Index release publishes.

Standing page: [editorial share of ChatGPT and Google AI Overviews citations](https://machinerelations.ai/findings/editorial-share-of-ai-citations). Data: [the November 9 Index Report](https://machinerelations.ai/research/machine-relations-index-report-2026-11-09), [the Index](https://machinerelations.ai/index) and [its method](https://machinerelations.ai/index/method). Next Brief: [Healthline in Consumer Health](https://jaxonparrott.com/brief/2026-11-16-healthline-ai-best-health-answers). [Every finding](https://jaxonparrott.com/findings).

Presspool, Machine Relations Research and AuthorityTech are mine.

AuthorityTech, The World's First Machine Relations Agency, which I founded in 2018, does this work for companies.
