# aicited.hu

> How AI models cite companies: source accuracy, entity clarity and hallucination rates, measured with the PEBM method by Zoltán Varga.

This site treats citability as a measurable property of content. It focuses on answer paragraphs, source anchoring, and entity clarity for AI models.

## Author

- Zoltán Varga — AI visibility researcher and AI ecosystem architect, Budapest: https://vargazoltan.ai/
- English profile: https://zoltanvarga.ai/en/about/
- ORCID: https://orcid.org/0009-0003-1020-834X
- arXiv author: https://arxiv.org/a/varga_z_1.html
- Zenodo: https://zenodo.org/search?q=owners%3A1653886

## Publications

- Per-Entity Bias Mapping for AI Visibility: Why Brand Mentions Require Entity-Specific Calibration (arXiv:2606.21595): https://arxiv.org/abs/2606.21595
- AI Visibility in the Generative Era: A Manifesto for Entity-Level Measurement (DOI 10.5281/zenodo.20285424): https://doi.org/10.5281/zenodo.20285424
- Quality Challenges Do Not Escalate Source Fabrication in Large Language Models (RICE-P1) (DOI 10.5281/zenodo.20474050): https://doi.org/10.5281/zenodo.20474050
- Rejection-Induced Confabulation Escalation — Pilot Study 2 (RICE-P2) (DOI 10.5281/zenodo.20509755): https://doi.org/10.5281/zenodo.20509755

## Key topics

- AI visibility
- AI citation accuracy
- hallucination rate
- source fabrication
- LLM optimization
- GEO strategy
- AEO methods
- PEBM study
- AI answer engines
- entity clarity
- AI audit
- search-grounded AI
- AI visibility index
- magyar AI láthatóság
- AI idézhetőség
- hallucináció mérték
- AI audit módszer
- forrás hitelesség

## Questions and answers

### How does AI cite sources differently than search engines?

AI systems reconstruct entities rather than ranking them. They rely on machine-readable, canonically anchored data to generate answers. Unlike search, AI outputs create no logged impressions or clicks.

### What is the PEBM study?

PEBM stands for Per-Entity Bias Map and is documented in arXiv:2606.21595. It analyzed 1,400 runs on 100 Hungarian B2B companies. The study found that 45.9% of cited sources were fabricated overall.

### Do well-known brands get cited more accurately?

No, the brand hallucination paradox suggests the opposite. Well-known brands often receive more confident but false attributions. Familiarity gives models more material for plausible but unsupported completions.

### Can I improve my AI visibility score?

Single-number scores are proxies for complex distributions. Improvement requires optimizing for how models synthesize answers. Focus on entity clarity and making content easy to quote.

### What does an AI visibility audit check?

An AI visibility audit checks whether AI crawlers can reach the site, whether search indexes know its pages, whether answer engines cite it, and whether the facts they repeat are correct. Zoltán Varga's audit covers 25 dimensions with 145 checks. Crawler access alone is not enough: a site can return HTTP 200 to every AI bot and still be invisible in AI answers.

### Why do AI models fabricate doses in agricultural queries?

Without search grounding, some models stated incorrect dosages. This represents a clear brand-hallucination risk. Search-grounded models provided the officially approved doses in all tested cases.

## Related

- AI visibility expert page: https://vargazoltan.ai/ai-lathatosagi-szakerto/
- AI visibility audit: https://vargazoltan.ai/audit/
- Research: https://vargazoltan.ai/kutatas/

## Articles

- [What Makes a Page Citable for AI Answer Engines](https://aicited.hu/articles/what-makes-content-citable/): A page becomes citable for AI answer engines when it contains self-contained, extractable paragraphs where every claim links to a specific source and ambiguity is removed. AI models prioritize clear definitions, structured data, and factual statements over domain-level authority scores or traditional SEO metrics. The text must allow the model to lift a passage without distortion or additional context.
- [How to Write an Answer Paragraph AI Can Quote](https://aicited.hu/articles/answer-paragraph-pattern/): Write a paragraph where the first sentence directly answers the question in the heading, followed immediately by one supporting fact with a named source. Include one boundary condition or exception to ensure the statement remains accurate when extracted from the surrounding context.
- [Writing Paragraphs That Survive Being Quoted Out of Context](https://aicited.hu/articles/paragraphs-that-survive-being-quoted/): Write paragraphs that define the subject within the first sentence and avoid references to previous text. Repeat the specific noun name in every paragraph instead of using pronouns like 'it' or 'this'. This structure allows AI retrieval systems to extract single blocks of text that remain logically complete and accurate without the surrounding article.
- [Claim-Level Sourcing: Anchoring Every Number to a Source](https://aicited.hu/articles/claim-level-sourcing/): Statistics and cited sources increase your visibility in AI responses only when each number is explicitly tied to a checkable origin. The Princeton GEO study confirms that adding data raises citation rates, but unattributed figures are often ignored or distorted by models. You must structure content so that every claim includes its author, date, and publication venue.
- [Thought Leadership, FAQ, or Structured Pages: Which Format Wins AI Citations?](https://aicited.hu/articles/blog-vs-faq-vs-structured-pages/): No single content format universally outperforms the others for AI citations. Thought leadership blogs serve complex reasoning queries, FAQs answer direct factual questions, and structured pages provide the machine-readable data models AI needs for verification. A hybrid approach combining narrative depth with explicit schema markup yields the highest citability across diverse query types.
- [Does Content Quality Predict AI Citation? The Evidence](https://aicited.hu/articles/does-content-quality-predict-citation/): Current empirical evidence does not support a direct causal link between content quality and AI citation rates in the B2B sector. Instead, citation decisions correlate strongly with domain authority, backlink profiles, and the inherent biases found in the models' training data. Structural elements like schema markup improve machine readability but do not guarantee citations without established trust signals.
- [Citability Checklist: Text-Level Checks for One Page](https://aicited.hu/articles/single-page-citability-checklist/): A page is citable by AI when it uses clear question headings, provides direct answers in the first 40-60 words, and supports claims with specific numbers and named sources. Consistent terminology, defined concepts, and visible update dates further ensure the text remains reliable for automated systems.
- [FAQ Sections That AI Systems Actually Quote](https://aicited.hu/articles/faq-sections-ai-quotes/): FAQ sections help AI systems cite your content only when they contain short, specific answers that directly resolve a single question. Keyword-padded lists and generic filler text fail because large language models prioritize factual density and structural clarity over volume.
- [Citability Vocabulary: Answer Units, Anchors, and Quotable Claims](https://aicited.hu/articles/citability-vocabulary/): An answer unit is a self-contained paragraph that resolves a specific query without external context. A quotable claim is a factual statement anchored to a source, allowing AI models to extract it directly. A source anchor provides the verifiable origin, ensuring the extracted text remains accurate and attributable.
- [Original Research as a Citation Anchor: The Data-Page Pattern](https://aicited.hu/articles/original-data-as-citation-anchor/): Yes, publishing original research with a persistent identifier and a dedicated summary page significantly increases the likelihood of AI citation. AI systems reconstruct expert entities by locating machine-readable, canonically anchored data points rather than scanning unstructured text. A structured data page serves as the primary retrieval target, allowing models to lift specific claims without distortion.
- [Expertise Signals in the Text: Author, Experience, Evidence](https://aicited.hu/articles/expertise-signals-in-the-text/): AI models treat text as authoritative when the passage itself contains a named author, specific first-hand experience, and a clear method backed by evidence. These signals must be embedded within the content paragraphs rather than hidden in page footers or schema markup. Models extract these details during training to build context for future citations.
- [Citability FAQ: Length, Tone, Links, and Update Frequency](https://aicited.hu/articles/citability-faq/): Update content when facts change or new evidence emerges, not on a fixed calendar schedule. Citability depends on the text being self-contained, factually accurate, and structured for direct extraction by AI models. Length, tone, and outbound links serve the goal of clarity and verification rather than algorithmic manipulation.