identity.json Explained: How Bing Reads Your Site's Identity for AI Search (2026 Guide)

identity.json is a JSON file at your site's root that declares who owns the site — business name, description, contact, social profiles — in machine-readable, Schema.org-aligned form. One vendor claims "Bing specifically fetches identity.json," but Microsoft publishes no documentation confirming it. No crawler is documented to consume it, and a UK survey of 1,500 sites found zero using it.

That gap between the claim and the evidence is exactly what this guide sorts out. Below: what identity.json actually is, who created the idea, how it compares to ai-catalog.json and llms.txt, what is verified versus what is still unknown, and step-by-step instructions for adding one to your site — plus an honest answer on whether it's worth your time in 2026 for a typical website owner.

What identity.json is (and who made it)

Think of it this way: robots.txt tells crawlers what they may access. identity.json would tell AI systems who you are — one canonical, machine-readable statement of publisher identity. Instead of an AI model piecing together your business name from your homepage headline, your About page, and a directory listing, the file hands it the facts directly: official name, what you do, how to contact you, where else you exist online.

Now the important part: no single company invented identity.json, and no standards body has ratified it. It grew out of the 2026 wave of "AI discovery files" — the same Cambrian explosion that produced llms.txt, ai.txt, agents.md, and ai-catalog.json. Two independent anchors give the idea its shape:

• The AI Visibility Definition (365i). A terminology project by the UK agency 365i — version 1.3.2, dated January 2026, with a draft ETSI technical specification — codifies identity.json as ADF-006: "Canonical structured identity data," sitting alongside llms.txt (ADF-001) and ai.txt (ADF-004) in a numbered catalog of AI discovery files. You can read the full term list in the AI Visibility Definition repository.
• An open community spec. A community "ai-discovery-files" specification on GitHub documents identity.json at version 1.0.0: a Schema.org-aligned JSON document served at the site root, with required fields like name, url, type, description, services, contact points, and a last-updated date.

Tool vendors then productized it — most visibly aiwebpageseo.com, which bundles an identity.json generator with its llms.txt and ai.txt tools. And that vendor is also the source of the Bing association you see repeated around the web: its homepage states that "Bing specifically fetches identity.json to understand who a site belongs to."

I went looking for the Microsoft documentation behind that sentence and found none — no learn.microsoft.com page, no Bing Webmaster blog post, nothing in Bing's published crawler or webmaster guidance mentioning an identity.json file. So the fair summary is: real file convention, real draft specs, but the Bing connection is one vendor's marketing claim, not documented behavior. Keep that distinction in mind for everything that follows.

identity.json vs ai-catalog.json vs llms.txt

The AI-discovery stack is getting crowded, and these files answer different questions. Here is how they compare:

The one-line version: identity.json is "who," ai-catalog.json is "what it offers," llms.txt is "what it says," ai.txt is "what's allowed," and robots.txt is "what's accessible." They are complementary, not competing. A site preparing seriously for the agent era would eventually publish all of them — but their maturity levels are wildly different, which matters for where you spend your time.

What is verified vs what is still unclear

This space rewards skepticism, so here is the honest ledger.

Verified

• The format is documented in at least two independent places. The community draft spec (v1.0.0) and 365i's AI Visibility Definition (ADF-006) describe the same idea: a Schema.org Organization-aligned JSON file with fields like name, url, type, description, services, contact points, sameAs profiles, and a last-updated date.
• The conventional location is the site root: yoursite.com/identity.json. Unlike ai-catalog.json, it has no .well-known convention — the drafts put it next to robots.txt at the root, served as application/json with an HTTP 200.
• Real-world adoption is effectively zero. A 2026 UK survey of roughly 1,500–2,000 domains — conducted by the same vendor that sells an identity.json generator — found zero sites serving one. That is a measurement, not an insult: the file is genuinely rare.
• The Bing claim exists as a marketing statement. aiwebpageseo's homepage says Bing fetches the file. What doesn't exist is any Microsoft documentation backing it up.
• Bing's actual, documented AI-search publisher tooling is elsewhere. Microsoft's real work in this area lives in Bing Webmaster Tools — the AI Performance dashboard (public preview, February 2026) showing which pages get cited in Copilot and Bing AI answers — and page-level controls like the data-nosnippet attribute (October 2025). None of it involves identity.json.

Still unclear

• Whether any crawler — Bingbot included — requests /identity.json today. Nobody has published crawler logs or documentation showing it. Until someone does, assume the audience for this file is future tooling, not current bots.
• Which schema would win if the idea takes off. The community draft, the vendor generators, and the terminology definitions agree on the concept but no authority has blessed one field list. Early files may need rewriting later.
• Whether it will ever influence citations or visibility in AI answers. There is no mechanism today connecting this file to how Copilot, ChatGPT, or AI Overviews choose sources. Identity signals that matter now — consistent Organization markup, real author bylines, verifiable contact info — work through channels crawlers already read.

How to add one to your site today

Adding the file takes about 15 minutes. Here is the whole job:

Step 1: Gather the facts

Write down the canonical version of your identity: official business name, legal name if different, one to three factual sentences describing what you do, your primary contact email, founding date, and the URLs of your real social and directory profiles. Keep every claim factual and checkable — this file is a trust signal, and inflated language in a trust signal does the opposite of its job.

Step 2: Write the JSON

There is no official schema to copy, so treat the example below as illustrative — it follows the field list from the community draft spec (name, url, type, description, services, contactPoints, sameAs, last-updated metadata). Replace every value with your own:

{
  "name": "Example Tools Co.",
  "url": "https://example.com",
  "type": "Corporation",
  "description": "A free suite of AI and SEO utilities for website owners.",
  "services": [
    "robots.txt generator",
    "llms.txt generator"
  ],
  "contactPoints": [
    {
      "type": "support",
      "email": "support@example.com"
    }
  ],
  "sameAs": [
    "https://www.linkedin.com/company/example-tools",
    "https://x.com/exampletools"
  ],
  "metadata": {
    "lastUpdated": "2026-10-12",
    "note": "Illustrative structure based on the community draft; no official schema exists."
  }
}

Step 3: Validate it

Paste the file into any JSON validator and confirm it parses — a single trailing comma silently breaks every consumer. Confirm it is served with the Content-Type: application/json header.

Step 4: Publish at the root

Upload it to https://yoursite.com/identity.json and check it returns HTTP 200:

curl -I https://yoursite.com/identity.json

You want to see 200 OK and content-type: application/json. If your server serves unknown extensions as downloads or 404s, add the MIME mapping for .json first.

Step 5: Don't stop at the file

This is the step most guides skip. The identity channel that search engines actually read today is JSON-LD Organization markup on your pages — put the same facts (name, url, logo, sameAs, contactPoint) in a script type="application/ld+json" block on your homepage. Google and Bing consume that right now; identity.json has no documented consumer yet. Do both, but do the JSON-LD first. Then write your llms.txt — that is the discovery file with a live audience.

Should you bother in 2026? Honest take

Yes, if you have 15 spare minutes. The file costs nothing, breaks nothing, and the exercise forces you to nail down one canonical version of your identity facts — which improves your About page, your structured data, and your directory listings anyway. If a consumer ever appears, you're already there. Early, cheap bets are how you win emerging standards.

No, if you expect results. Do not publish identity.json expecting traffic, citations, or ranking changes. There is no documented consumer, measured adoption is zero, and the Bing connection is an unconfirmed vendor claim. Anyone selling you "Bing identity optimization" around this file is selling the claim, not the evidence.

Do these first instead — every one of them has a live audience today:

1. Publish llms.txt pointing AI systems at your best content.
2. Add JSON-LD Organization markup with matching identity facts on your homepage.
3. Keep your name, contact, and profiles consistent across your site, social accounts, and directories — inconsistency is what actually confuses entity resolution.
4. Verify the site in Bing Webmaster Tools and watch the AI Performance report for real citation data.
5. Write clear author bylines and an About page a human (or model) can verify.

Who benefits most from identity.json specifically? Brands fighting impersonation — if scraper sites or copycats muddy who the real you is, a canonical, self-published identity file is one more piece of "this is the real us" evidence, cheap to add. For everyone else, it's a lottery ticket: nearly free, nearly certainly not paying out this year, but correctly priced at zero.

FAQ

Is identity.json an official Bing or Microsoft standard?

No. It is a community and vendor-driven convention — documented in a community draft spec and in 365i's AI Visibility Definition (ADF-006) — with no Microsoft documentation behind it. The "Bing fetches identity.json" line comes from a tool vendor's marketing page, not from Bing.

Where exactly do I put identity.json?

At your domain root: https://yoursite.com/identity.json, served as application/json with an HTTP 200 status. Unlike ai-catalog.json, it has no .well-known convention in the current drafts.

Will identity.json help me rank in Bing or get cited in Copilot answers?

There is no evidence it affects rankings or citations. No crawler is documented to consume it. Bing's documented publisher controls for AI search are the AI Performance dashboard in Bing Webmaster Tools and page-level attributes like data-nosnippet — not this file.

Is identity.json the same as Organization schema markup?

Related, different channel. Organization JSON-LD on your pages is consumed by Google and Bing today and powers knowledge panels and entity understanding. identity.json is a standalone file with the same kind of facts but no documented consumer yet. Publish the JSON-LD first; the file is optional.

Should I pay for a tool that generates identity.json?

No — not for this file alone. It's a 15-minute hand-write using facts you already know, and every generator is guessing at an unofficial draft schema anyway. The value is in having correct, consistent identity data everywhere, not in the file itself.