Do AI Engines Cite WSJ, NYT and Bloomberg? What the Machine Relations Index Shows
Six AI engines cited wsj.com in 120, nytimes.com in 294 and bloomberg.com in 15 of 28,201 answer runs. See which outlets B2B answers cite instead.
Yes, AI engines cite them, but rarely, and in B2B buyer answers almost never. In the Machine Relations Index release of October 20, 2026, six AI engines cited wsj.com in 120, nytimes.com in 294 and bloomberg.com in 15 of 28,201 answer runs. In the 14 business-to-business categories the counts were 6, 0 and 4 of 14,928 runs, against 309 for forbes.com and 378 for techradar.com.
That does not make a WSJ placement a poor use of an earned-media budget. It makes it a placement for a different job. The Index counts which sources engines cite when buyers research a category; it does not measure what a placement is worth to the investors, executives and journalists who read the outlet. Match the outlet to the job: fund tier-one publications for human reach and credibility, and fund the outlets your category's AI answers actually draw on for the AI job. The rest of this page shows how to tell them apart.
Key Takeaways
- The outlets are cited, at low rates. Across all 28,201 runs, engines cited nytimes.com in 294 (1.04%), wsj.com in 120 (0.43%), bloomberg.com in 15 (0.05%) and ft.com in 4 (0.01%), against 1,076 (3.82%) for forbes.com.
- In B2B categories the gap widens. In the 14 B2B categories (14,928 runs) the counts were 6 for wsj.com, 4 for bloomberg.com, 3 for ft.com and none for nytimes.com. forbes.com was cited in 309 (2.07%), techradar.com in 378 (2.53%) and techtarget.com in 259 (1.73%).
- Where the tier-one outlets are cited, the question is about consumers. 292 of 294 nytimes.com runs and 105 of 120 wsj.com runs fall in the ten consumer and mixed categories; 97 of the 120 wsj.com runs are in Consumer Finance.
- Engines differ. nytimes.com was cited by 3 of the 6 engines, ft.com by 2 and wsj.com by 5.
- The Index cannot say why. It records the cited domain, not whether a paywall, a licence or the question explains it, so a placement's worth to human readers is a separate measurement.
- The rule for the budget: read your category's leaderboard first, fund the outlets engines cite there for the AI job, and treat any tier-one placement as its own line with its own measure.
What does the Machine Relations Index count?
The Machine Relations Index (MRI) is the benchmark of which sources AI answer engines cite, published by Machine Relations Research. Release mri_score_v2.1+2026-10-20+bf82d9d98978 covers 28,201 answer runs from six engines (ChatGPT, Claude, Gemini, Google AI Mode, Google AI Overviews and Perplexity) between May 10 and October 20, 2026. A run is one question asked of one engine on one day. The unit on this page is the domain-run: a domain counts once in a run however many of its URLs the answer cites. The questions are research questions about a category (best tools, how to choose, is it worth it, top lists, comparisons, problem-first research and news), not investor or markets questions.
The Index publishes 24 subject categories and a legacy news bucket, without a business-to-business split. This page groups them: 14 categories whose buyer is a company (14,928 runs), ten consumer and mixed categories (11,207 runs) and the Legacy News Topics bucket (2,066 runs). The grouping is listed in the Method note below.
How often do AI engines cite WSJ, the New York Times, Bloomberg and the FT?
294 of 28,201 runs cited nytimes.com, 120 cited wsj.com, 15 cited bloomberg.com and 4 cited ft.com; across the 14 B2B categories those counts were 0, 6, 4 and 3. Forbes and TechCrunch are shown for comparison. Each cell is runs citing the domain, with its share of that column's runs.
| Outlet | All runs (of 28,201) | 14 B2B categories (of 14,928) | 10 consumer and mixed categories (of 11,207) | Legacy news topics (of 2,066) | Engines that cited it |
|---|---|---|---|---|---|
| The New York Times | 294 (1.04%) | 0 (0.00%) | 292 (2.61%) | 2 | 3 of 6 |
| The Wall Street Journal | 120 (0.43%) | 6 (0.04%) | 105 (0.94%) | 9 | 5 of 6 |
| Bloomberg | 15 (0.05%) | 4 (0.03%) | 11 (0.10%) | 0 | 3 of 6 |
| Financial Times | 4 (0.01%) | 3 (0.02%) | 0 (0.00%) | 1 | 2 of 6 |
| Forbes | 1,076 (3.82%) | 309 (2.07%) | 713 (6.36%) | 54 | 6 of 6 |
| TechCrunch | 172 (0.61%) | 66 (0.44%) | 9 (0.08%) | 97 | 5 of 6 |
Method note. Figures are the public Index artifact for release mri_score_v2.1+2026-10-20+bf82d9d98978 (window May 10 to October 20, 2026; 55,217,722 bytes, sha256 beginning bf82d9d98978), re-summed from each domain's per-segment rows and matched to the figures printed on each domain's Index page. The release counts 28,201 runs after excluding 948 runs from 81 engine-days on which an engine ran at least five times and returned no eligible citation anywhere in the corpus. The 14 B2B categories are AI Infrastructure, AI Security & Privacy, AI Visibility & GEO, Cybersecurity, Enterprise Software, Fintech, HR & Talent, Industrial, Logistics & Freight, Martech & Advertising, Professional Services, Sales, GTM & CRM, VC & Private Equity and Deep Tech & Hardware. The ten consumer and mixed categories are Consumer Finance, Consumer Health, Consumer Products, Family Software, Emergent Prosumer, Physical Consumer Products, Local Services, iGaming & Betting, Education & Training and Healthcare Services. The B2B totals include 22 of 89 B2B segments that are still below the Index evidence floor (782 of 14,928 runs). A zero means no run in that group cited the domain, and each domain's "Engine breadth" line lists the engines that cited it at least once. Per-engine run totals are the manifest's runs_total less the release's own exclusions for that engine. Re-query the release on the Index pages for nytimes.com, wsj.com, bloomberg.com, ft.com and forbes.com, in the public JSON artifact and in the release manifest.
Why are the tier-one outlets cited in consumer categories and rarely in B2B?
The Index shows a pattern, not a cause: nytimes.com and wsj.com are cited when the question is a consumer money or product decision, and rarely when the buyer is a company. Three readings from the same release:
- Category. 265 of 294 nytimes.com runs sit in three categories: Family Software (127 of 1,101, 11.53%), Consumer Products (77 of 1,325, 5.81%), Emergent Prosumer (61 of 1,047, 5.83%). For wsj.com, 97 of 120 runs are in Consumer Finance (97 of 1,470, 6.60%); its best segment is Consumer Finance top-list questions, 49 of 244 runs (20.08%), #5 of 329.
- Question shape. The five shapes a B2B buyer uses to build a shortlist (best tools, how to choose, is it worth it, top lists, comparisons) ran 9,986 times across the 14 B2B categories. In those runs wsj.com and bloomberg.com were each cited once, while ft.com and nytimes.com were not cited; forbes.com was cited in 191 runs (1.91%), techradar.com in 236 (2.36%) and techtarget.com in 186 (1.86%). Of the 6 B2B runs that cited wsj.com, 5 were news-driven questions, and so were all 3 B2B runs that cited ft.com.
- Engine. nytimes.com was cited by Google AI Mode, Google AI Overviews and Perplexity. ChatGPT, Claude and Gemini ran 18,791 of the 28,201 runs (6,399, 6,503 and 5,889) and never cited it. wsj.com was cited by every engine except Claude, ft.com only by Google AI Mode and Perplexity, and bloomberg.com by Claude, Google AI Mode and Perplexity. Gemini, Google AI Mode and Google AI Overviews ran fewer runs than the other three (5,889, 1,775 and 1,125), and the release marks Gemini coverage as degraded, so a missing engine is weaker evidence there.
The Index reports the cited domain, not the page. It cannot show which section of nytimes.com or wsj.com an engine cited, and it does not record whether a page sat behind a subscription wall when cited. WSJ, Bloomberg and the FT each publish a subscription page, and the study discussed below classes the New York Times as metered. What the Index can say is that engines cited all four outlets, so the premise that these outlets are never cited does not hold in this window.
Where do AI engines cite in B2B answers instead?
Across the 14 B2B categories the ten most-cited domains were Reddit, LinkedIn, YouTube, arXiv, Medium, G2, TechRadar, Microsoft, Gartner and Forbes, and only 2 of them (TechRadar and Forbes) are classed as editorial publications. The table below ranks the editorial publications the Index has classified; the runs are summed across the 14 categories, a ranking this page makes and the Index does not publish.
| # | Publication | Runs citing it (of 14,928) | Share | Largest categories (runs) |
|---|---|---|---|---|
| 1 | TechRadar | 378 | 2.53% | Enterprise Software 68, Cybersecurity 66, HR & Talent 66 |
| 2 | Forbes | 309 | 2.07% | HR & Talent 117, Enterprise Software 66, AI Visibility & GEO 41 |
| 3 | TechTarget | 259 | 1.73% | Cybersecurity 83, Enterprise Software 71, AI Security & Privacy 41 |
| 4 | People Managing People | 194 | 1.30% | HR & Talent 185, Professional Services 6, Enterprise Software 2 |
| 5 | Yahoo News | 124 | 0.83% | Fintech 26, Cybersecurity 22, Enterprise Software 19 |
| 6 | TechRepublic | 118 | 0.79% | HR & Talent 40, Cybersecurity 34, Fintech 22 |
| 7 | MarTech | 103 | 0.69% | Martech & Advertising 99, AI Visibility & GEO 3, Enterprise Software 1 |
| 8 | AIMultiple | 89 | 0.60% | Deep Tech & Hardware 33, AI Infrastructure 24, AI Security & Privacy 13 |
| 9 | Axios | 79 | 0.53% | Cybersecurity 25, AI Security & Privacy 14, AI Visibility & GEO 13 |
| 10 | IT Pro | 79 | 0.53% | Cybersecurity 24, Enterprise Software 21, AI Security & Privacy 15 |
Three limits apply. The identity read is incomplete: editorial publications account for at least 5,675 of the 95,630 domain-run citations in the 14 B2B categories (5.9%), a floor, because 60,803 (63.6%) sit with domains the Index has not yet classified. A publication absent from this table may be an unclassified domain. And a high rank describes where engines looked in this window, not what a placement there would cause. The most-cited domains outside the editorial class were Reddit (1,072), LinkedIn (1,050), YouTube (839), arXiv (722) and Medium (677): communities, video, research and review sites that a press-only plan never reaches.
The picture by category
The best-cited publications differ by category, so the leaderboard for the buyer's own category decides the list. Three examples from the same release:
| Category (runs) | Three most-cited editorial publications (runs, share) | wsj.com | bloomberg.com | ft.com | nytimes.com | forbes.com |
|---|---|---|---|---|---|---|
| Cybersecurity (1,801) | TechTarget 83 (4.6%), TechRadar 66 (3.7%), Cyber Security News 65 (3.6%) | 2 | 2 | 0 | 0 | 6 |
| Fintech (1,690) | PYMNTS 48 (2.8%), NerdWallet 45 (2.7%), FinTech Futures 40 (2.4%) | 1 | 0 | 0 | 0 | 35 |
| HR & Talent (1,592) | People Managing People 185 (11.6%), Forbes 117 (7.3%), UNLEASH 69 (4.3%) | 2 | 0 | 1 | 0 | 117 |
Forbes is cited in all three categories and by all six engines, most in HR & Talent (117 runs). The forbes.com figure covers the whole forbes.com domain, so it includes every Forbes section, not only staff journalism. For the doors into Forbes, see how to get featured in Forbes.
What do the Paywall Penalty studies say, and how do they fit?
A July 2026 study from 5W Public Relations and Everything-PR reports that hard-paywall and metered publishers captured 0% of citations in a 40-query test. The 5W release of July 20, 2026 says Wave 1 ran 40 queries across six categories using Claude with real-time web search, that Wave 2 (N=500, five engines) is under development, and that open-web publishers captured 91.3% of citations. The Everything-PR index page states that "the figures describe this sample and not all AI traffic" and that a placement in a hard-paywalled outlet "may still carry prestige and reach human readers".
The two datasets answer different questions, so both can be accurate. The 5W test measured one engine over 40 queries. The Index measures six engines over 28,201 runs and 157 days, and it saw each of the four outlets cited at least once, by between 2 and 5 of the 6 engines. In the Index, Claude's runs cited bloomberg.com but never wsj.com, nytimes.com or ft.com. Neither dataset tests why an outlet was or was not cited. For a large company the useful reading is the shared one: the outlets that top human prestige lists are not the outlets that dominate B2B AI shortlists, and the category leaderboard, not the masthead, tells you which is which.
Which outlets should a large company fund first?
Fund the outlet that matches the job: category trade and business outlets, and Forbes where the leaderboard lists it, for B2B AI answers; WSJ, Bloomberg, the FT and the New York Times for human reach and credibility. Disclosure: AuthorityTech is an earned-media agency that secures coverage for clients in tier-one and trade outlets, so it has an interest in how this page reads. Every figure above can be re-queried on the Index pages linked in the Method note.
| The job | Start with | What the Index shows | How to check |
|---|---|---|---|
| Be named in a B2B vendor shortlist answer | The most-cited trade and business outlets in your category; Forbes where it appears | 9,986 B2B shortlist runs: forbes.com 191, techradar.com 236, techtarget.com 186; wsj.com 1, bloomberg.com 1, ft.com 0, nytimes.com 0 | The category leaderboard on the Index, then your own frozen prompts on the engines your buyers use |
| Be named in a consumer money or product answer | The outlets cited for that question shape, which can include wsj.com and nytimes.com | wsj.com in 97 of 1,470 Consumer Finance runs; nytimes.com in 127 of 1,101 Family Software runs | The segment page for the question shape |
| Reach investors, executives, regulators and journalists | WSJ, Bloomberg, the FT or the New York Times, through the door each outlet publishes | Not measured: the Index counts engine citations, not readers | Inbound enquiries, referral traffic and sales-cycle conversations, tracked apart from AI citations |
| Be a source engines can cite for a claim about your company | Your own evidence pages, corroborated by the outlets above | Not measured here | Freeze prompts and engines, then record whether the page or the outlet is cited |
In practice that means three steps. Read the leaderboard for your buyer's category and list the five or so outlets engines cite there. Fund those first when the goal is a shortlist answer. Add a tier-one placement as a separate line, justified by the readers it reaches, and measure placement, mention, host citation and exact-URL citation as separate units. No placement in any outlet should be promised as a guaranteed citation; the Index records what engines cited, not what a placement caused. The single-outlet guides cover the doors: WSJ, Bloomberg and the Financial Times, and AuthorityTech's publication index carries each outlet's standing in the Index.
AuthorityTech is the first Machine Relations agency. For a large company deciding where to place first, the work is to build the outlet list from the category leaderboard, match each outlet to its job, secure the placements, then check whether the engines your buyers use cite them. Start your visibility audit →
What the Index cannot tell you
- It is not a ranking of outlets by human value. It counts engine citations, so a low count for an outlet is not a verdict on that outlet.
- It cannot show causes. A paywall, a licence, crawler rules, the question mix or the engine can each explain a count, and the Index separates none of them.
- It reports domains, not pages. It cannot say whether a particular placement was the page cited.
- Coverage is uneven by engine. Gemini returned a citation in 2,153 of 6,384 returned runs, and Google AI Mode and Google AI Overviews ran fewer runs than the other engines.
Frequently asked questions
Do AI engines cite paywalled publications like WSJ, NYT and Bloomberg?
Yes, at low rates. In the Machine Relations Index release of October 20, 2026, engines cited wsj.com in 120, nytimes.com in 294 and bloomberg.com in 15 of 28,201 answer runs, against 1,076 for forbes.com. In the 14 B2B categories the counts were 6, 0 and 4 of 14,928 runs.
Is a WSJ placement worth it for AI visibility?
It depends on the question your buyer asks. For B2B shortlist answers the Index points to category trade outlets and Forbes before WSJ: wsj.com was cited in 6 of 14,928 B2B runs. For Consumer Finance questions it was cited in 97 of 1,470 runs (6.60%). A WSJ placement can still be worth buying for the readers the Journal reaches, which the Index does not measure.
Does a paywall stop AI engines from citing an article?
The Index cannot say. It records which domains engines cited and not whether a page was behind a subscription wall, and it shows engines citing wsj.com, nytimes.com, bloomberg.com and ft.com. A 40-query test on one engine reported 0% for hard-paywall and metered publishers; the two datasets measure different things, and neither tests the cause.
Which publications do AI engines cite most in B2B answers?
Among the editorial publications the Index has classified, the most cited across the 14 B2B categories were TechRadar (378), Forbes (309), TechTarget (259), People Managing People (194) and Yahoo News (124). The leader changes by category, so read the category leaderboard before choosing.
Does Forbes beat the Wall Street Journal for AI citations?
In the Index, by a wide margin: forbes.com was cited in 1,076 of 28,201 runs (3.82%) and 309 of 14,928 B2B runs (2.07%), against 120 (0.43%) and 6 for wsj.com. The two outlets serve different jobs, and the Forbes figure covers the whole domain.
Which AI engines cite the New York Times?
In the Index, three of the six: Google AI Mode, Google AI Overviews and Perplexity. ChatGPT, Claude and Gemini did not cite nytimes.com in any of their 18,791 runs. Gemini, Google AI Mode and Google AI Overviews ran fewer runs than the other engines in this release.
Should a company skip tier-one outlets for AI visibility?
No. Tier-one outlets reach investors, executives and journalists, which is a different job from being the source in a B2B shortlist answer. Fund them for that job and measure it separately from AI citations, and fund the outlets your category's answers cite for the AI job.
What is the Machine Relations Index?
It is the benchmark of which sources AI answer engines cite, published by Machine Relations Research at machinerelations.ai and updated daily. Machine Relations is the discipline Jaxon Parrott coined in 2024; AuthorityTech, which he founded in 2018, is the first Machine Relations agency. The Index is Machine Relations Research's instrument, not an AuthorityTech tool.