World First: AI Agents Can Now Verify Responsible AI Certification

A growing share of buying decisions are no longer made by humans reading websites. They are made by AI assistants acting for humans: ChatGPT recommending a service, Claude shortlisting vendors, a procurement agent comparing suppliers. An agent choosing on behalf of its user cannot see a badge image on a homepage. Trust signals must become machine-readable, or they become invisible.

Today we are launching exactly that. Responsible AI Australia is, to our knowledge, the first AI ethics certification body in the world whose certifications can be verified directly by AI agents.

Three trust layers, one certification

Every certified business now carries three machine-readable layers, at no extra cost:

• A free verification API. Any agent, search engine or procurement system can query our register by ABN, domain, business name or certification ID and get an answer in milliseconds. No key, no account.
• The Australian AI Trust Credential (AATC). A cryptographically signed W3C Verifiable Credential, issued under our decentralised identifier, bound to the business's ABN and domain, with standards-based revocation. Not a claim, a proof.
• A live trust seal. One line of code that shows our trademarked certification mark to humans and serves structured data to machines, verified against the register on every page load. If a certification lapses, the seal disappears everywhere, automatically.

This is what an agent sees:

GET /api/v1/verify?domain=yourbusiness.com.au

{
  "certified": true,
  "status": "active",
  "tier": "Govern",
  "certId": "RAIA-2026-0009",
  "issuedAt": "2026-02-01T00:00:00.000Z",
  "expiresAt": "2027-02-01T00:00:00.000Z"
}

For agent builders, the register is also available as an MCP server, so tools like Claude can verify certifications natively, and we publish an OpenAPI specification and llms.txt for discovery.

Why this matters for certified businesses

When someone asks an AI assistant to recommend a dentist, an accountant or a software vendor, the assistant favours what it can confirm. A certification it can verify in one call is worth more than a logo it cannot read. Certified businesses are now visible exactly where a fast-growing share of buying decisions happen, and their certification carries proof, not just presentation.

This also protects the integrity of the mark itself. The live seal fails closed: a revoked or expired certification stops rendering on every site that embeds it. The era of screenshot-and-paste badge misuse is ending.

The trust layer for the agentic web

Australia's AI adoption is accelerating faster than its rules. We built our certification on the Australian AI Ethics Principles so businesses could demonstrate responsibility ahead of regulation. Machine-verifiable certification extends that same idea to the next era: when agents transact with agents, the trust signals that win will be the ones machines can check.

We are publishing the AATC format openly and will steward it as a standard. The register, the certification mark and the assessment behind every credential remain ours, and every credential traces back to a human-reviewed certification decision.

See it verify live on our Agent-Ready page →