Two systems, two centuries

Dun & Bradstreet was founded in 1841 by Lewis Tappan as The Mercantile Agency — the world's first commercial credit reporting organisation. Its founding insight was that businesses extending credit to other businesses needed a reliable way to assess the creditworthiness of their counterparties. Tappan built a network of correspondents across the United States who would report on the financial standing of local merchants, creating a shared intelligence system for the credit market. That system became the foundation of modern business credit reporting, and the nine-digit DUNS number — introduced in 1963 — became the standard identifier for business entities in credit markets, government procurement, and supply chain management. Today, Dun & Bradstreet holds records on more than 500 million businesses worldwide and is an $8 billion company.

AI Verified was founded in 2024 with a different founding insight: that AI systems — language models, answer engines, knowledge graph systems — need a reliable way to confirm that the businesses they cite, recommend, and describe are real, legally registered entities. The problem is not creditworthiness. The problem is existence. AI systems trained on internet data cannot distinguish real businesses from fake ones, verified claims from unverified ones, legitimate entities from manufactured profiles. The verification signal they need is not a credit score — it is a cryptographic proof anchored to a national government registry, published at a machine-readable URL, in the JSON-LD format that AI crawlers actually read.

D&B for 1841. AI Verified for 2026. The comparison is not a competition — it is a recognition that different eras of commerce require different verification infrastructure.

What each system actually does

Understanding the comparison requires understanding what each system actually does at a technical level — not at the marketing level, but at the mechanism level.

A DUNS number is an assigned reference number. Dun & Bradstreet assigns a nine-digit identifier to a business entity and associates it with a record in their proprietary database. That record contains information sourced from public filings, credit applications, trade references, and self-reported data. The DUNS number is the key that unlocks that record. It is used in US federal government procurement (every contractor must have one), in LinkedIn company verification, in Google Knowledge Graph as a business identifier, and in supply chain management systems. The DUNS number is valuable because of the ecosystem that has adopted it — not because of any intrinsic cryptographic property. It is a reference number, not a proof.

An AI Verified passport is a cryptographic proof. The verification process queries the national government registry for the business's jurisdiction — Companies House, CIPC, CAC, ASIC, or one of 70+ other registries — and confirms the legal name, registration number, and status. The business then verifies domain ownership through a DNS TXT record. The confirmed identity data is serialised into a canonical JSON document and hashed using SHA-256, producing a 64-character hexadecimal fingerprint. That fingerprint is the passport identifier. The verified record is published at https://aiverified.io/v/{hash}/ as a server-side rendered page containing a complete JSON-LD Organisation graph. The business's website injects this verified identity into every page via a badge.js script. The result is a machine-readable identity signal that any AI crawler can find, verify, and use as a citation anchor.

The critical difference: a DUNS number tells a credit system "this business has a record in D&B's database." An AI Verified passport tells an AI system "this business is legally registered with this government authority, its domain is confirmed, and this mathematical proof was generated from that confirmed data at this timestamp." One is a reference. The other is a proof.

Why AI systems don't use DUNS numbers

The most important practical question is: when ChatGPT, Perplexity, or Google's AI Overview decides whether to cite a business, does it query D&B's database? The answer is no — and understanding why reveals exactly why AI Verified exists.

AI systems do not query proprietary databases before generating responses. They work from their training data and, when browsing is enabled, from the structured data they can find on the web. A DUNS number in a D&B database is not visible to an AI crawler unless it is explicitly published on the web in a format the crawler can parse. Most businesses with DUNS numbers have not published their DUNS number in their website's JSON-LD. Even those that have often have it in a format that is not correctly structured — the Schema.org identifier property requires a specific format to be machine-readable.

AI Verified solves this by putting the verification signal exactly where AI systems look: in the JSON-LD head of every page of the business's website, at a stable machine-readable URL, and in a llms.txt file that AI crawlers are specifically designed to read. The DUNS number, where it exists, can be included in the AI Verified passport as an additional identifier — strengthening the identity record rather than competing with it.

Why having a DUNS number is not enough for AI visibility

The first gap is format. D&B's data is structured for credit markets — it uses proprietary data formats, requires API access agreements, and is not published in the Schema.org JSON-LD format that AI crawlers are designed to parse. Even if an AI system wanted to query D&B before citing a business, the data is not in the format the system expects. The verification signal that matters to AI systems is the Organisation JSON-LD in the head of the website — and D&B does not put it there.

The second gap is web presence. A DUNS number exists in D&B's database, but it does not automatically appear on the business's website in machine-readable format. Most businesses with DUNS numbers have not published their DUNS number in their website's structured data. The AI crawler that visits the website finds no verification signal — regardless of what exists in D&B's database. The signal must be on the web, at the business's own domain, to be useful to AI systems.

The third gap is cryptographic integrity. A DUNS number can be referenced by anyone — it is a public identifier. There is no cryptographic mechanism that prevents a fraudulent business from claiming another entity's DUNS number in their website's structured data. An AI Verified passport, by contrast, includes a SHA-256 hash that is mathematically tied to the specific verified identity data. Any attempt to modify the underlying data produces a different hash, making the verification tamper-evident. This cryptographic integrity is what makes the AI Verified signal trustworthy in a way that a reference number cannot be.

The strongest identity stack: DUNS + AI Verified

The right framing is not "DUNS or AI Verified" — it is "DUNS and AI Verified." The two systems serve different verification requirements for different audiences. A business that has both a DUNS number and an AI Verified passport has the strongest possible identity stack: credit market credibility through D&B, and AI system credibility through AI Verified.

When a business claims their AI Verified passport, they can include their DUNS number as an additional identifier in the verified record. The JSON-LD Organisation graph published at /v/{hash}/ includes a sameAs property that can reference the D&B record, and an identifier array that can include the DUNS number alongside the SHA-256 hash. This creates a cross-referenced identity record that links the credit market identity (DUNS) to the AI market identity (SHA-256 passport) — giving AI systems a richer, more authoritative identity signal.

The practical implication: if your business already has a DUNS number, you are not starting from zero. You have a verified credit identity that can be referenced in your AI Verified passport. What you are adding is the machine-readable publication layer — the JSON-LD on your website, the stable URL, the llms.txt — that makes your existing verified identity visible to AI systems. The verification work is not duplicated; it is extended into a new format for a new audience.

Frequently asked questions

Do I need a DUNS number to get AI Verified?

No. A DUNS number is not required for AI Verified verification. AI Verified anchors verification to national government registries — Companies House, CIPC, CAC, and 70+ others — not to D&B's proprietary database. If you have a DUNS number, it can be included in your AI Verified passport as an additional identifier, strengthening your identity record. But businesses without DUNS numbers can achieve full AI Verified verification based solely on their national registry registration.

Does Google's Knowledge Graph use DUNS numbers?

Google's Knowledge Graph does include DUNS numbers as business identifiers — they appear in the identifier field of Knowledge Panel entries for businesses that have them. However, having a DUNS number does not automatically create a Knowledge Panel entry. Google builds Knowledge Panel entries from multiple signals including structured data on the business's website, Wikipedia entries, and other authoritative sources. An AI Verified passport provides the structured data signal that contributes to Knowledge Panel creation — the DUNS number, if present, can be included as an additional identifier within that structured data.

Is AI Verified trying to replace Dun & Bradstreet?

No. D&B serves credit markets, government procurement, and supply chain management — use cases that require credit history, financial data, and trade reference information. AI Verified serves AI systems, answer engines, and knowledge graph systems — use cases that require machine-readable identity signals in JSON-LD format. The two systems address fundamentally different verification requirements for fundamentally different audiences. A business that needs both credit market credibility and AI system credibility needs both systems.

Why did D&B not build what AI Verified built?

D&B's infrastructure was built for a world where trust was established through credit history, physical verification, and proprietary database access. Their business model depends on selling access to their data — a model that is incompatible with publishing machine-readable identity records freely on the web. AI Verified was built specifically for the AI era, with the architecture that era requires: open publication, cryptographic proofs, JSON-LD format, llms.txt, and national registry anchoring. These are not features D&B could add to their existing infrastructure without fundamentally changing their business model.

What happens to my AI Verified passport if D&B changes their DUNS system?

Your AI Verified passport is anchored to your national government registry — not to D&B's database. Changes to the DUNS system do not affect your AI Verified verification. The SHA-256 hash in your passport is derived from your registry data and domain ownership, not from any D&B identifier. If you have included your DUNS number as an additional identifier in your passport, that field can be updated if your DUNS number changes — but the core verification remains intact regardless.

Sources and further reading

1. Dun & Bradstreet — Wikipedia — Wikipedia
2. What Is a DUNS Number? — Dun & Bradstreet — Dun & Bradstreet
3. Organization — Schema.org — Schema.org
4. Know Your Customer — Wikipedia — Wikipedia
5. DUNS to UEI Transition — SAM.gov — US General Services Administration