Publishers and Developers FAQ
Official information about OAI-SearchBot, GPTBot, referral tracking and agent accessibility.
Public web discovery and separate crawler controls
OpenAI, ChatGPT Search and Website Access
OpenAI distinguishes search discovery, potential model-training access and agent interaction. Those are related but separate publisher decisions.
The crawler associated with ChatGPT search discovery
OpenAI advises publishers who want public content discovered, surfaced and cited in ChatGPT search not to block OAI-SearchBot. Crawl permission remains only one requirement; the server, CDN and security configuration must also allow the request.
OpenAI also explains that ChatGPT referral links can include a source parameter, giving publishers a way to identify click-through traffic in analytics.
A separate control for potential training
OpenAI identifies GPTBot separately from OAI-SearchBot. A publisher can make different choices about search visibility and potential model-training use when the platform provides distinct user agents.
The Hidden Radius package follows the requested open-crawler policy, but organizations should make that choice deliberately rather than assuming every OpenAI user agent performs the same role.
A directive must be accessible to be read
OpenAI notes that if a publisher wants a page excluded from certain search surfacing, a noindex directive may be relevant. The crawler must be permitted to access the page to read that directive. This illustrates a broader technical rule: a blocking mechanism can prevent the system from seeing the instruction intended to control indexing.
Interactive websites need understandable controls
OpenAI’s developer guidance connects agent performance with accessible page structure and ARIA semantics. Descriptive roles, labels and states help an agent understand what a button, menu or form control does.
Hidden Radius pairs that platform guidance with W3C standards: use native semantic HTML where possible, add ARIA where needed, provide accessible names, support keyboard operation and require confirmation for consequential actions.
Permission is not selection
- Allowing OAI-SearchBot does not guarantee crawling frequency.
- Crawling does not guarantee indexing or citation.
- A citation does not guarantee that every generated claim is fully supported.
- Referral traffic does not capture every form of visibility or later customer action.
- The same public page can behave differently across product modes and user contexts.
Primary material behind this analysis
ARIA Authoring Practices Guide
Patterns for accessible names, roles, states, landmarks, controls and keyboard operation.
Open the original source at W3C Web Accessibility Initiative →
Continue through the discovery stack
Crawling, robots.txt and Indexing Explained
A technical guide to discovery, crawler access, robots.txt, infrastructure blocks, directives, canonical URLs, sitemaps and indexing.
AI Agents, Tool Use and Machine-Operable Websites
A guide to agent loops, browser interaction, accessible controls, confirmation, prompt injection, permissions and safe task completion.
Search and AI Visibility Measurement Framework
A layered measurement framework for crawler access, indexing, entity recognition, retrieval, citations, representation accuracy and business outcomes.
Search and AI Research Library
A research hub connecting NIST, Stanford HAI, OECD, Google, OpenAI, Perplexity, Microsoft, IETF, W3C, Schema.org and original GEO, RAG and REG sources.