Press Releases Earn 0.04% of AI Citations. The Editorial Coverage They Generate Earns 81%.

I have spent nearly a decade placing brands in exactly the publications that press releases are designed to reach. AuthorityTech has generated thousands of media placements across every major tier. I know how the wire works, what it costs, and what it does.

Here is what it does not do: get cited by AI.

BuzzStream and Citation Labs analyzed 4 million AI citations across ChatGPT, Google AI Mode, Google AI Overviews, and Google Gemini. They ran 3,600 prompts across 10 industries and tracked every source the engines pulled from.

Press releases distributed through syndication channels like Yahoo and MSN accounted for 0.32% of news citations. That is 0.04% of the entire dataset. Direct citations from wire services like PR Newswire made up 0.21% of total citations.

Original editorial content accounted for 81%.

Read those numbers again. Then look at your last wire invoice.

What the Wire Service Industry Is Selling

Wire services know AI search is the new terrain. They are marketing directly into it.

ACCESS Newswire offers an "AI Visibility Checklist" for press releases. eReleases published a guide positioning releases as tools for AI search visibility. Business Wire wrote about optimizing releases for answer engine discovery. TrackMyVisibility built an entire guide around the premise that press releases can be optimized for AI engines.

The pitch is consistent: write your press release the right way, distribute it on the right wire, and AI search will cite it.

The data says something else.

4 Million Citations and a 0.04% Reality

BuzzStream used XOFU, a citation monitoring tool from Citation Labs, to track where AI platforms pull their sources. The methodology: 3,600 prompts, 10 industries, four major AI platforms, one week of collection.

The breakdown is blunt. Within news citations, syndicated content overall (articles republished through MSN and Yahoo networks) accounted for 6.2% of news citations and 0.9% of total citations. Press releases through those same channels were a fraction of that fraction: 0.32% of news, 0.04% of total.

Wire services appeared most often on exploratory and informational prompts. Even there, they only reached 0.37%.

Original editorial content did not merely outperform syndicated press releases. It occupied a different universe. 81% of all news citations in the dataset. The split held across every prompt type BuzzStream tested.

For evaluative prompts like "Is Sony better than Bose?", news citations hit 18% of all citations. The editorial content that showed up most: head-to-head comparisons and cost analysis from Reuters, CNBC, and CNET.

Those are not publications you reach by distributing a press release. Those are publications that write about you because you earned it.

The Wire Hierarchy Does Not Match the Citation Hierarchy

Everything PR ran the first independent wire-by-wire citation audit in June 2026. Twenty-four prompts, two engines (Claude and Google AI Overviews), five wire services.

Their headline finding cuts through the entire wire sales pitch: "The press release is the input. The editorial rewrite is what the AI engines surface as the answer."

The spend hierarchy is inverted. Companies paying $2,000 to $3,000 per release on PR Newswire and Business Wire often get zero citation surface inside Claude and Google AI Overviews. Companies paying $400 to $800 on GlobeNewswire, Newsfile, and ACCESS Newswire sometimes get direct domain citation.

GlobeNewswire took 31.3% citation share among wires. Business Wire and PR Newswire tied at the next tier: 18.8% and 16.7%. But the critical finding is not which wire wins. It is that 21% of prompts returned zero wire surface across every engine and every wire tested. Every one of those zero-citation prompts was either a mega-cap consumer tech brand (Apple, Meta, Microsoft, Nvidia), a regulated mid-cap, or a generic category-level question.

When CNBC, Variety, Quartz, MacRumors, or BioSpace process a wire release into editorial coverage, the engines cite the editorial layer. Not the wire. When Morningstar, Stocktitan, or FinancialContent syndicate the same release with the wire attribution preserved, the engines cite the wire's name. But editorial rewrites are what the AI engine surfaces as the answer.

The press release is the catalyst. The editorial rewrite is the product.

The One Exception Worth Understanding

ChatGPT treats owned newsroom content differently than every other engine.

In BuzzStream's dataset, internal press releases and newsroom content published on company-owned domains accounted for 18% of ChatGPT's citations. On Google's AI platforms, that number dropped to roughly 3%.

The examples tell the story. When prompted about Iberdrola's role in renewables, ChatGPT cited a press release from Iberdrola's own website. When asked about Target's products, ChatGPT cited a 2015 press release from Target's corporate domain.

This is the one scenario where a press release can become the cited source: when it lives on your domain, not a wire. And when the prompt is specifically about your company.

That is a narrow window. It is also a structural argument for publishing every release on your own newsroom before you distribute it anywhere else. If ChatGPT is going to cite a press release, it is going to cite yours, from your domain. Not the version that ended up on Yahoo Finance through a $2,500 wire send.

What This Actually Means for Your Communications Budget

5WPR synthesized findings from more than 680 million tracked AI citations across ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, and Google AI Mode. The finding that should restructure every communications budget: brand search volume is the strongest known predictor of AI citation likelihood, with a 0.334 correlation. That is materially stronger than backlinks.

Brand-building activity compounds directly into AI visibility. That is not a theory. That is what 680 million citations say.

And only 11% of domains are cited by both ChatGPT and Perplexity. A single press release strategy, distributed once through a single wire, reaching a single syndication network, will not succeed across the AI surface. The surfaces are structurally different.

Zero-click search behavior rose from 56% of queries in 2024 to 69% by May 2025. AI models cite between 2 and 7 domains per response, according to SellersCommerce. The funnel narrowed. The wire is not inside it.

The math is straightforward. If you are spending $2,000 to $3,000 per release on wire distribution and that release generates zero editorial pickup in publications AI engines actually cite, your wire spend is not an AI visibility investment. It is an expensive announcement. Stop calling it a strategy.

The Move: Stop Treating the Press Release as the Finished Product

I coined Machine Relations because the old system stopped working and nobody was naming the new one. This data makes the point cleaner than any argument I could write.

The press release has always been raw material. In the old system, the raw material was valuable because journalists processed it into coverage, and that coverage reached human readers. The value chain was simple: wire release, journalist picks it up, article runs, human reads it.

The new value chain is identical in structure and completely different in who reads the output. Wire release, journalist picks it up, article runs, AI engine reads it, AI engine cites it, human reads the AI answer.

The press release's job did not change. It still exists to trigger editorial coverage. What changed is the downstream reader. And that downstream reader, the AI engine, cites the editorial layer 81% of the time and the wire layer 0.04% of the time.

Here is the move. Stop spending more on distribution. Start spending more on what happens after distribution. Get placed in the publications that AI engines actually pull from. Structure the earned coverage so the machine can extract a clean, attributable claim. Build entity clarity across enough independent surfaces that the engine does not have to guess who you are.

That is not a press release optimization problem. That is citation architecture.

Go check your last quarter of wire spend. Then check how many of those releases generated editorial coverage in a publication that ChatGPT, Perplexity, or Google AI Mode actually cites. If you do not know the answer, you do not have a measurement problem. You have an architecture problem.

The press release is the raw material. Stop treating it as the finished product.

FAQ

Do press releases help with AI search visibility?

Press releases distributed through wire services account for 0.04% of AI citations in BuzzStream's 4-million-citation dataset. The editorial coverage those releases generate accounts for 81% of news citations. The press release is valuable as a catalyst for earned media, not as a directly cited source. Publishing on your owned newsroom before distributing on a wire gives ChatGPT the best chance to cite the release directly, since ChatGPT cites company newsroom content at 18% in the same dataset.

Which wire service gets the most AI citations?

GlobeNewswire leads with 31.3% citation share among wire services, followed by Business Wire at 18.8% and PR Newswire at 16.7%, according to Everything PR's 2026 Wire Service Citation Audit. The audit found no single dominant wire: each service wins a different vertical. The larger finding is that AI engines cite the editorial rewrite of a release far more often than the wire source itself.

What content type gets cited most by AI search engines?

Original editorial content accounts for 81% of news citations across ChatGPT, Google AI Overviews, Google AI Mode, and Google Gemini, according to BuzzStream's analysis of 4 million citations. Within evaluative prompts, head-to-head comparisons and cost analysis from outlets like Reuters, CNBC, and CNET dominate. Brand-owned newsroom content accounts for 18% of ChatGPT citations but only 3% on Google platforms.

What is citation architecture?

Citation architecture is the system of entity clarity, source-layer authority, structured extractability, and cross-domain corroboration that determines whether AI engines cite a brand. It was defined by Jaxon Parrott, founder of AuthorityTech, as part of Machine Relations. Citation architecture is what separates brands that get cited from brands that get ignored, regardless of press release volume or wire spend.

Should companies stop using press release distribution services?

No. The press release is still the catalyst that triggers editorial coverage. The correction is in expectations and investment allocation. 5WPR found that brand search volume is the strongest predictor of AI citation likelihood (0.334 correlation). Press releases that generate editorial coverage in publications AI engines trust compound into citation surface. Press releases that sit on a wire syndication network without generating editorial pickup do not. The budget should shift toward ensuring the editorial coverage happens after the wire sends, not toward optimizing the wire distribution itself.