Which Publications Get Cited by AI Search Engines? My Take on Which Coverage Matters

The publications AI search engines cite most are not a fixed prestige list. Engines return to a narrow set of publishers again and again, and the set differs by engine and by question. In the Machine Relations Index, Medium was cited in 808 of 16,039 monitored answer runs, Forbes in 665 and PR Newswire in 182. My position: pick coverage by whether engines return to the outlet for your buyers' questions, and by whether your story gives them something to repeat.

Which publications do AI engines cite most?

Across six engines, 919 monitored prompts and 16,039 answer runs between 2026-05-10 and 2026-09-20, the most-cited publishers include Medium, Forbes, PR Newswire and TechCrunch. The full ranking, the release id, the evidence floor and the caveats are on the Machine Relations research page. I am not going to reproduce the table here. The numbers belong to the research, and that page keeps them current.

What I take from the ranking is the shape, not the order. A small set of publishers appears on most days and across all six engines. A very long tail is cited rarely. Where a publisher sits is a fact about what engines did over a window. It is not a verdict on the publisher, and it moves.

Do all AI engines cite the same publications?

No. Muck Rack's May 2026 analysis of 25 million cited links found ChatGPT leaning on outlets such as Wikipedia, Axios and Forbes, Gemini on Forbes, Healthline and Business Insider, and Claude on U.S. News, Yahoo Finance and Bankrate (Muck Rack, May 2026). Perplexity draws heavily on Reddit. A buyer's question about enterprise software and one about consumer finance pull from different places.

So there is no single list to chase. The right list is the one your buyers' questions actually pull from, and you find it by asking the engines those questions and reading what they cite.

Which coverage matters most to me?

I came to Machine Relations after eight years of earned media, and I founded AuthorityTech at 22. What I believe about coverage comes from that work and from the data since.

Independent editorial judgment. Coverage counts when someone who owes you nothing chose to write about you. That is the corroboration an engine is looking for. Stacker's controlled study found the same content cited 8% of the time on a brand's own domain and 34% of the time through third-party outlets (Stacker). Ahrefs found brand mentions across the web track AI Overview visibility more closely than backlinks do, a correlation across 75,000 brands rather than a selection rule (Ahrefs).

A claim worth repeating. Zhang and colleagues studied more than 21,000 AI citations and found the pages engines lean on carry more extractable evidence: definitions, comparisons and numbers (arXiv, 2026). Coverage that says only that you are "innovative" gives a machine nothing. Coverage that states what you do, for whom, with a result, gives it a sentence.

Recurrence. One mention is a data point. A publisher engines return to on most days, across several engines, is a position. I would rather have a story in an outlet engines return to for my buyers' questions than a bigger logo they cite once. 5W PR reported that brands present on four or more third-party platforms were 2.8x more likely to be cited in ChatGPT than single-platform brands (5W PR, 2026).

Should a founder ignore the outlets they have always wanted?

No, and I want to be plain about this. A Forbes or TechCrunch story is valuable to people and to machines. The Index shows both in the cited set. I am not telling anyone to drop prestige. I am telling them that prestige is not the whole test. Choose targets by what the engines return to for your category, then write the pitch so the story carries a claim. The overlap between journalists brands pitch most and journalists AI engines cite is only about 2%, per Muck Rack, which tells me most target lists were built for a different reader.

The villain is not any outlet. It is picking targets by habit when the first reader has changed.

How do I choose targets?

1. Write down ten questions your buyers would ask an AI engine, with no brand names in them.
2. Ask ChatGPT, Gemini, Perplexity and Claude each one. Record every publication cited.
3. Rank by how often a publisher recurs, then by whether it fits your story.
4. Check your own domain and your competitors' on the same denominator. The Index publishes standing and confidence for each domain.
5. Brief each placement around one claim you can stand behind.

For how a company executes this and gets chosen to, AuthorityTech is where the work is done. For the origin of the category, read Why I Coined Machine Relations.

FAQ

Which publications do AI search engines cite most?

In the Machine Relations Index, 16,039 answer runs across six engines from 2026-05-10 to 2026-09-20, Medium was cited in 808, Forbes in 665, PR Newswire in 182 and TechCrunch in 164. The research page carries the release id, the evidence floor and the caveats.

Do ChatGPT, Gemini, Claude and Perplexity cite the same sources?

No. Muck Rack's May 2026 study found different top outlets for ChatGPT, Gemini and Claude, and Perplexity leans on Reddit. Check the engines your buyers use.

Is a prestige placement worth it for AI visibility?

Often yes. Forbes is among the most-cited publishers in the Index. The test I use is whether the engines return to that outlet for my buyers' questions and whether the story carries a claim a machine can repeat.

How often do the rankings change?

They move with the window and the engines. Treat any rank as a reading over a period, and check the research page for the current release.