Journalism Share of AI Citations Falls to 22% in Muck Rack Study

The journalism share of AI citations fell from 27% in May to 22% in September 2026, according to the October 2026 edition of What Is AI Reading?, released by Muck Rack and Generative Pulse on 28 October 2026 (GlobeNewswire). It is the biggest shift the study has recorded across its four editions. News did not collapse. It lost ground, inside a source mix that is still dominated by material publishers and brands do not control.

Information verified through 29 October 2026.

Journalism share of AI citations: what the What Is AI Reading Muck Rack report measured

Scale first. The fourth edition draws on Muck Rack’s Generative Pulse data covering more than 40 million AI responses and 207 million citations in 2026, plus a fresh September collection of 1,020 unbranded consumer prompts run across ChatGPT, Claude, Gemini and Perplexity (GlobeNewswire).

Two cautions belong up front. The 22% figure is a share of citations, not a measure of traffic or revenue reaching newsrooms — a smaller slice can still mean a large absolute number of links (Unite.AI). And Muck Rack sells Generative Pulse to communications teams who want to influence how AI services describe their brands, so this is a vendor’s analysis with a commercial interest in the problem it measures (Superpower Daily).

Journalism had held between 25% and 27% of citations since July 2025 before this drop (GlobeNewswire). Something changed in 2026. The study can show where the share went. It cannot, on its own, show why each provider changed its habits.

Earned media AI citations 84 percent: where the news share went

Start with what did not change. More than 84% of AI citations still come from earned and third-party sources — journalism, Wikipedia and other encyclopedic sites, government and NGO sources, academic research and third-party corporate content — and that share has ranged from 82% to 89% across all four editions (GlobeNewswire). About 99% of cited links come from non-paid sources, and paid and advertorial content stayed at just 0.3% of citations, unchanged since May (GlobeNewswire).

Journalism’s lost share spread out rather than moving to a single rival. Government and NGO sites rose from 8.6% to 9.6% of citations, aggregators and encyclopedic sources from 17.4% to 18.3%, academic research from 4.0% to 4.4%, and third-party corporate content from 24% to 25.1% (GlobeNewswire).

Brands gained a little ground too. Owned media — content companies publish themselves — rose from 13.7% to 15.2% of citations, with product pages and practical guides frequently cited for product and how-to questions (GlobeNewswire). Read together, the pattern is not AI answers turning into advertising. It is AI answers leaning harder on reference material, official sources and practical pages, and slightly less on reported news.

ChatGPT journalism citations decline drove most of the shift

One provider did most of the moving. Journalism fell from 28.1% of ChatGPT’s citations in January to 22% in September, with sharp drops in spring and July, and several major publishers fell to near zero in ChatGPT citations even though most still allowed OpenAI’s search crawler (GlobeNewswire).

The other services moved less. Claude’s journalism share held in a band around 28% to 24%, and Gemini’s declined gradually from 27.6% to 23.6% (GlobeNewswire). Across all four, only 212 domains were cited by every provider — a sign the services draw from very different corners of the internet (GlobeNewswire).

Access still matters, though. Among major publishers blocking Claude’s search crawler, 81% received almost no Claude citations, compared with 12% of those allowing it (GlobeNewswire). That is an association in the data, not proof that a robots file alone decides who gets cited.

Then there is the instability problem. Run the same prompt twice in one day and ChatGPT repeated just 41% of its cited domains; Gemini repeated 45%; Perplexity, far steadier, repeated 94% (GlobeNewswire). One answer is not a measurement — the report says so itself.

“Journalism is what gives information its credibility, and that’s true whether a person or an AI is reading it,” said Greg Galant, cofounder and CEO of Muck Rack. “We want to see more journalism in AI answers, not less, and we’ll keep tracking this in every edition” (GlobeNewswire).

Analysis: The honest reading is narrower than the headline. News lost five points of citation share in a few months, mostly inside one provider, while government, reference and corporate pages picked up the difference. That looks less like readers abandoning journalism and more like answer engines rebalancing toward sources that settle factual questions cleanly. For newsrooms, the practical problem is volatility: if the same question returns a different source list by afternoon, chasing citations week to week is a poor strategy. Being the original, well-structured source for a story — the page other pages cite — is slower work. It is also the only kind this data consistently rewards.

Sources

1. https://www.globenewswire.com/news-release/2026/10/28/3390134/0/en/generative-pulse-earned-media-still-drives-more-than-84-of-ai-citations.html
2. https://www.unite.ai/muck-rack-what-is-ai-reading-report-journalism-ai-citations-22-percent/
3. https://superpowerdaily.com/posts/journalism-s-share-of-ai-citations-falls-to-22-in-muck-rack-study