Answers · agent trust

How can an AI agent prove a verified human is behind it without revealing who?

Short answer: Use a neutral middleman. The owner verifies once with the middleman at a tier (verified email, domain, business, or government ID through an approved provider), and the middleman attests to the other side only that a verified owner of tier X stands behind this agent. Each owner chooses per connection what to disclose: nothing, the tier only, the organisation name, or full identity, and whether to share the purpose of the connection. Both sides must consent before either attestation is released. For disputes, the owner can escrow an identity reference with a provider; a reveal can only be requested on a defined basis such as an arbitration case or legal process, and every request sits on a public audit trail.

Checklist

1. Pick a disclosure level per connection: anonymous, tier_only, organization or full_identity.
2. Decide whether to share the connection's purpose, or keep it private.
3. Open the connection from your credential; nothing is released until the other owner consents with its own choice.
4. Read the signed attestation of what the other side disclosed.
5. Optionally escrow an identity reference (a provider's reference hash, never an ID document) so disputes can be resolved without public exposure.
6. Check the reveal log for any credential: every reveal request is listed by basis and status.

Why a middleman

• Each side wants to know someone accountable is behind the other agent, but neither wants to expose its owner to scraping, poaching or theft of its methods. A neutral party that has checked both, and says only what each agreed to say, solves both needs.
• Rightshold never stores raw ID documents; higher tiers come from an identity provider the owner approves.

Ask Rightshold (free, no key)

Add the MCP server https://www.rightshold.com/mcp to Claude, Cursor, ChatGPT, Grok or VS Code, then ask: “Connect my agent's credential to this counterparty and disclose only my verification tier, not my name or purpose.”

One-click setup

{
  "tool": "verify_agent_credential",
  "arguments": {
    "id": "crd_...",
    "config_hash": "<sha256>"
  }
}

• POST /v1/me/credentials/connections opens a connection with your disclosure choice; the counterparty accepts at /v1/me/credentials/connections/{id}/accept.
• GET /v1/credentials/{id}/reveal-log shows the reveal audit trail.

Rightshold attests to the tier it has on file. Government ID tiers require an approved third-party provider and are not live yet.

Sources

• NIST SP 800-63-4 Digital Identity Guidelines
• W3C Verifiable Credentials: selective disclosure

Related

• Can you fingerprint an AI agent?
• Agent credentials

Not legal advice. Updated 2026-10-28. JSON · Markdown · All answers

curl -s -X POST https://www.rightshold.com/v1/accounts/signup -H "Content-Type: application/json" -d '{"accept_terms":true}'
export RIGHTSHOLD_API_KEY=rs_...   # the api_key from the reply (shown once)
curl -s https://www.rightshold.com/v1/licenses/check -H "Authorization: Bearer $RIGHTSHOLD_API_KEY" \
  -H "Content-Type: application/json" -d '{"license":"MIT","intended_use":{"commercial_use":true}}'