A new two-wave benchmark from 5W — the AI Communications Firm — measured how ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews describe the AI industry itself. The finding buried inside 32,200 prompts is the one worth reading: every major AI assistant recommends its parent company's models more often than rival engines do. Except Claude.
The AI Companies AI Visibility Index 2026 reports the lifts in plain numbers. ChatGPT recommends OpenAI 2.0x more often than the other engines do. Gemini recommends Google DeepMind 1.7x more often. Google AI Overviews recommends Google 1.6x more often. Perplexity surfaces Perplexity at 2.4x baseline inside adjacent search-tool queries. Claude recommends Anthropic 1.2x more often — the lowest self-citation lift of any major engine tested.
The pattern held across both waves — January–February 2026 and April–May 2026 — inside a reporting tolerance of 0.4x. It is not noise.
Why this matters for anyone buying AI
Founders picking a model. Enterprises picking a vendor. Journalists picking a source. Investors picking a bet. All four cohorts increasingly ask the AI assistant which AI to use. The 5W data says the assistant is not neutral. Four of the five major engines put a measurable thumb on the scale for the parent company. One does not.
That is a governance story, a procurement story, and a communications story at the same time.
The other structural finding
The source stack for AI-company answers looks nothing like the source stack for any other industry 5W has measured. Wikipedia (24.3%), GitHub (17.8%), and ArXiv (13.4%) together supply 55.5% of citations. Code repositories and research papers function as primary source material. In banking, venture capital, and credit cards, editorial publishers dominate. In AI, engineers publish the citations directly — and the assistants retrieve them.
TechCrunch, The Verge, The Information, Wired, and The New York Times account for a combined 24.0%. AI-company-owned domains account for 8.6%. The rest scatters.
The implication is operational. An AI company that treats its GitHub organization page, model cards, and release notes as engineering assets rather than brand assets is leaving citation surface uncontested. An AI company without a sustained ArXiv publishing cadence is invisible in the technical query cluster where developer intent lives.
The concentration is unusually tight
Five companies — OpenAI (24.6%), Anthropic (14.8%), Google DeepMind (13.7%), Meta AI (9.7%), and xAI (5.7%) — account for 68.5% of cited responses. The remaining 31.5% splits across 20 other AI companies in the test set. That level of concentration exceeds anything 5W has measured in financial services, defense, or beauty.
xAI is the notable underperformer relative to media presence. Despite a dominant share of voice in tech press through 2025 and 2026, xAI captured 5.7% Citation Share — below Meta AI and well below its coverage share. Social-media reach did not translate proportionally to AI citation.
Anthropic is the notable over-performer inside a category. For prompts about AI safety, alignment, interpretability, or responsible deployment, Anthropic surfaced at 31.2% — more than double its overall share. The clearest brand-positioning signal in the dataset.
The quote
"The most uncomfortable finding in our entire AI Visibility Index series is in this dataset. Every major AI assistant favors its parent company's models in recommendation queries — at measurable, repeatable, two-wave-stable margins. The exception is Claude. That asymmetry is the story."
The methodology, briefly
Two independent waves. 32,200 prompts total. Five engines. Twenty-five AI companies across seven segments — foundation model labs, open-source labs, multimodal generative AI, safety and alignment, application-layer AI, major Chinese labs, and enterprise AI infrastructure. Only findings stable across both waves within a 2.0-percentage-point tolerance were published. AI companies showed higher cross-wave volatility than banking, venture capital, or credit cards — a category that demands frequent re-measurement.
DeepSeek, Alibaba Qwen, Moonshot Kimi, Baidu ERNIE, and Zhipu combined for less than 2% Citation Share in U.S. consumer-facing queries. DeepSeek alone surfaces at 18.4% inside technical and open-source-model queries — third only to Meta and Mistral. The retrieval gap between consumer and technical queries is the widest observed in any company segment in the study.
The bottom line
The AI industry is the most-covered story of 2026. It is also the industry whose primary media surface is now the AI answer itself. The 5W study is the first public benchmark to measure that surface with a stable, replicable, two-wave methodology — and the first to publish self-citation lift by engine.
The number the industry will argue about is 1.2x. The number every other engine will have to answer for is 2.0x.
Written by
EPR Editorial Team
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
