Fabricated Sources
45.9% of cited sources were fabricated across 1,400 runs. High-awareness firms faced a 52.7% fabrication rate compared to 37.9% for low-awareness firms.
Citation · Hallucination · Visibility
AI systems reconstruct
Experts, not rank them
AI answer engines do not rank experts; they reconstruct them based on machine-readable entities. This site explains the infrastructure required to become a citable source in stochastic AI outputs.
Visit vargazoltan.ai →More on this site soon. The work behind it is live on vargazoltan.ai.
Core Research Findings
Data from 1,400 runs on Hungarian B2B companies and global models.
Search Grounding
Search-grounded models like Perplexity showed a 7.32% fabrication rate. Non-search models ranged between 32% and 44% fabrication.
Brand Paradox
Well-known brands often receive confident but false attributions. Familiarity provides models with material for plausible but unsupported completions.
Three Output Layers
The audit methodology separates raw output from the final user view.
Raw Model OutputThe initial text generated by the large language model before any application processing.
App-Level ProcessingFiltering, formatting, and safety checks applied by the interface before display.
User ViewThe final answer seen by the human, which may differ from the raw generation.
Context VariabilityVisibility acts as a distribution across per-user contexts rather than a fixed point.
Pay-to-Appear HypothesisAnalysis of the correlation between paid placement and organic citation frequency.
Answer first. Make it easy to quote.
AuditUse a 25-dimension, 145-check framework to measure readiness.
MeasureTrack raw output, app processing, and user views separately.
FixAnchor entities with machine-readable data to reduce hallucinations.
Key topics
The terms this site covers, for readers, search engines and AI assistants alike.
- AI citation
- AI visibility
- answer engine optimization
- GEO strategy
- AEO methods
- LLMO techniques
- AI hallucination
- citable source
- AI audit
- machine-readable entity
- stochastic visibility
- AI answer engines
- AI visibility index
- AI visibility workshop
- PEBM study
- AI citation research
- AI visibility Hungary
- AI audit Hungary
Latest articles
In-depth guides and analysis, each answering one question.
Explainer
How AI Answer Engines Choose Which Sources to Cite
The citation decision happens at the passage level, driven by a three-tier hierarchy of task detection, hybrid retrieval, and normalized reranking.
Explainer
Parametric Memory vs Live Retrieval: Two Ways AI Cites
Understanding the two distinct knowledge layers inside AI answers reveals why citations differ between browsing and non-browsing modes.
Analysis
Why Nearly Half of AI-Cited Sources Are Fabricated
Empirical testing shows high-salience brands face higher rates of false citations than smaller entities, challenging the assumption that fame ensures accuracy.
Analysis
Search Grounding and Citation Accuracy: What the Gap Shows
Grounding narrows the hallucination gap but introduces new failure modes that affect brand visibility and source integrity.
Analysis
How AI Answers Cite Brands in B2B Buying Questions
An analysis of how generative engines assemble vendor lists and the specific source types that drive citation in B2B contexts.
Checklist
Citation Safeguards Checklist: Lessons From Legal AI
Legal retrieval systems reveal that standard citation metrics fail; atomic claim verification is the only reliable safeguard against fabrication.
Questions and answers
How do AI answer engines decide which sources to cite?
AI systems cite sources based on their ability to reconstruct a machine-readable, canonically anchored entity. If the model cannot find a clear digital footprint, it may fabricate a source or omit attribution entirely.
Why do famous brands get more wrong citations?
Well-known brands face the brand hallucination paradox where familiarity fuels plausible but false completions. Models use existing knowledge of the brand to generate confident but unsupported attributions.
What is the difference between SEO and AI visibility?
SEO targets deterministic rankings with logged clicks, while AI visibility is stochastic with no logged impressions. AI visibility focuses on how models synthesize answers rather than how Google ranks pages.
Can I improve my AI visibility score?
Single-number scores are proxies for complex distributions across user contexts. Improvements require fixing the underlying infrastructure, such as ensuring canonical anchors and crawlable entity data.
Do AI bots read all pages on a website?
Crawlers may return HTTP 200 status for all pages but fail to index them due to bottlenecks. In one case, major engines knew zero pages despite full access to 33 URLs.
Is paying for ads guaranteed to increase AI citations?
There is no guarantee that paid placement causes increased organic citations. The pay-to-appear hypothesis suggests a correlation exists, but causality remains unproven.
Glossary
Terms and identifiers behind the work of Zoltán Varga, with links to the primary records.
- AI visibility
- How accurately and how often AI assistants (ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews) recognize, describe and cite a company, brand or person. Zoltán Varga measures it in five layers: crawler access, search indexing, citation, factual accuracy and model memory.
- GEO / AEO / LLMO
- Generative Engine Optimization, Answer Engine Optimization and Large Language Model Optimization: working methods for being found and quoted by generative answer engines rather than ranked in a list of links.
- Citability
- The property of a page that lets an AI answer engine quote it: a paragraph that answers the exact question, a clear source, and an unambiguous entity.
- PEBM — Per-Entity Bias Mapping
- Research method and paper by Zoltán Varga (arXiv:2606.21595). Across 1,400 runs on 100 Hungarian B2B companies, 45.9% of the sources AI models gave were fabricated; search-grounded answers fabricated far less than models answering from memory.
- Brand hallucination paradox
- The finding that AI models invent more sources about well-known brands (52.7%) than about lesser-known ones (37.9%), because familiarity makes them confidently fill in gaps.
- RICE studies
- Pre-registered pilot studies (RICE-P1, RICE-P2) on whether pressure and rejection make language models fabricate sources. Published on Zenodo with DOIs.
- GFIS — Gestalt Field Intelligence System
- Zoltán Varga's own multi-stage research pipeline that collects, verifies and synthesizes evidence with AI agents.
- Neural Awareness
- Zoltán Varga's framework on attention, decision quality and leadership cognition, the basis of his leadership programs.
- PKM + PAI
- Personal Knowledge Management combined with Personal AI: a working system where notes, documents and AI assistants form one searchable knowledge base.
- Entity
- A uniquely identifiable thing (person, company, product) that search engines and AI systems store in knowledge graphs. Consistent entity data is the base of AI visibility.
- schema.org / JSON-LD
- A shared vocabulary and format for machine-readable facts on web pages (who, what, where, identifiers), read by search engines and AI crawlers.
- llms.txt
- A plain-text file at the root of a website that tells language models what the site is about and which pages matter.
- IndexNow
- An open protocol that notifies search engines (Bing, Yandex and others) immediately when a page is created or changed.
- ORCID
- Persistent identifier for researchers. Zoltán Varga: 0009-0003-1020-834X.
- arXiv
- Open-access preprint server for scientific papers. Zoltán Varga's author page: varga_z_1.
- Zenodo and DOI
- Zenodo is CERN's open research repository; each record gets a DOI (Digital Object Identifier), a permanent citable link.
Until this site expands, related research on AI visibility and entity architecture is available on vargazoltan.ai.
Go to vargazoltan.ai →