Context
Chen and Koudas report a 31,200-answer controlled audit of how declared and network location influence sources cited and businesses recommended by tested ChatGPT and Gemini configurations.
What changed
The October 16, 2026 preprint systematically compared 300 questions, 19 application-declared ChatGPT locations, five network locations across ChatGPT and Gemini, and repeated runs. It found different geographic signals were used by the tested systems, with resulting citation and recommendation differences.
Why it matters
AI visibility and GEO audits that use a single geographic origin can mischaracterize brand and source exposure. Declared location, network country and run-to-run variation need to be recorded as separate measurement variables.
Visibility OS interpretation
Visibility OS interprets this as evidence that geographic context is a first-class audit dimension for AI source selection and retrieval. The reported platform differences are conditional on tested settings, not permanent universal behavior or a guaranteed optimization lever.
What remains uncertain
How stable are these patterns after model, search backend and geolocation changes? How do logged-in personalization, language, local intent and device location interact with declared and network locations?