The cryptographic certificate standard for AI agents

SSL for AI Agents

AIGP-Σ is the open protocol for certifying AI agents — post-quantum identity, cryptographic scope enforcement, and a public registry anyone can verify.

Issues ML-DSA certificates for AI agents
Maintains the public agent registry
Provides business & reseller solutions

ML-DSA

Post-Quantum Signatures

< 1s

Certificate Issuance

99.9%

Registry Uptime

Open

Public Protocol

What is AIGP-Σ?

AIGP-Σ (AI Governance Protocol — Sigma) is a universal cryptographic protocol for autonomous agent governance. It defines how AI agents prove identity, declare authorized scope, and remain verifiable throughout their lifecycle — built on NIST post-quantum standards (ML-DSA, SHA3-512) with RISC Zero zero-knowledge proofs.

AIGP-Σ = The Protocol

Issues, registers, and standardizes. Open standard layer.

AIGodfather = The Platform

Monitoring, kill switches, compliance. Built on AIGP-Σ.

The Protocol — 5 Whitepaper Pillars

WP-01

Protocol Specification

Core cryptographic protocol built on NIST post-quantum standards (ML-DSA-65, SHA3-512) with RISC Zero zero-knowledge proofs. Four-layer architecture: Authority, Action Chain, IoT Delegation, Audit.

WP-02

The Kill Switch

Cryptographic agent revocation with four HALT levels: Soft HALT, Hard HALT (k-of-n threshold), Fleet HALT, and Emergency HALT (≤1s propagation). Post-quantum signed, publicly verifiable.

WP-03

Certificate Authority

SSL/TLS for AI agents. Every credential is ML-DSA signed, listed in a public append-only registry with optional Bitcoin anchoring — structurally equivalent to Certificate Transparency logs.

WP-04

Agentic Payments

The missing identity layer for autonomous payments. Works with x402 (Coinbase), Mastercard Agent Pay, and Stripe Agentic — certifying which agent is paying and whether it’s authorized.

WP-05

Multi-Agent Trust Chains

Cryptographic delegation with scope subset enforcement. Orchestrator → sub-agent hierarchies, HALT cascade propagation, cross-organization trust — no shadow delegations possible.

Cryptographic Foundation

ML-DSA-65

NIST FIPS 204 — post-quantum signatures (Category 3)

SHA3-512

NIST FIPS 202 — hash chains, Merkle roots

RISC Zero ZK-STARKs

Zero-knowledge proofs for authority verification

Bitcoin Anchoring

OP_RETURN timestamps — immutable proof of issuance

Industry Use Cases

AIGP-Σ is industry-agnostic. Only three parameters vary per deployment: Godfather authority structure (k-of-n), action criticality levels, and checkpoint frequency.

AI Governance

Agent action logging, scope enforcement, tamper-evident audit trail

Healthcare

Diagnostic agent certification, emergency halt for misdiagnosis

Finance & Payments

Trading bot identity, payment scope limits, x402/Stripe integration

Supply Chain

Logistics agent verification, cross-org trust between partners

Voting Systems

Election agent certification with 4-of-7 governance threshold

DAO Governance

On-chain governance agent trust, fleet halt for systemic issues

Legal & Contracts

Contract execution agent audit, immutable action chain evidence

IoT & Edge

Gateway-delegated HMAC signing for resource-constrained devices (~512B RAM)

What an AIGP-Σ certificate provides

Every certificate is an ML-DSA signed credential listed in the public registry. Here’s what you get when you certify your agent.

Certificate Issuance

Post-quantum ML-DSA certificates that prove agent identity, authorized scope, and model integrity.

Public Registry

Append-only public registry. Anyone can verify any agent’s status, scope, and revocation history.

Scope Enforcement

Every certificate cryptographically encodes what an agent is allowed to do. No overrides possible.

Reseller & Business API

White-label API for organizations and resellers. Issue certificates at scale under your own brand.

Built on AIGP-Σ

Need a Kill Switch?
That’s AIGodfather.

AIGodfather is the production platform built on AIGP-Σ. It adds real-time agent monitoring, behavior anomaly detection, human-in-the-loop approvals, and a cryptographic kill switch — all powered by AIGP-Σ certificates.

Every AIGodfather subscription includes a free AIGP-Σ certificate. Monitor hundreds of agents across fintech, healthcare, and e-commerce — with full EU AI Act compliance built in.

Visit AIGodfather

Emergency Kill Switch

Cryptographic HALT. Stop any agent in <1s, post-quantum signed.

Free Certificate Included

Every AIGodfather plan includes a free AIGP-Σ certificate.

Real-time Monitoring

Every agent action logged, classified by AI, risk-scored live.

Human-in-the-Loop

Require human approval before critical agent actions execute.

EU AI Act Compliance

Built-in compliance wizard, readiness scoring, regulatory export.

Security Scanning

OWASP LLM Top 10: prompt injection, tool abuse, memory poisoning.

Three steps to certified AI

Go from zero to verified in minutes, not weeks.

01

Register Your Agent

Sign up, define your agent’s identity, scope, and authorized actions.

02

Get Certified

Receive a post-quantum ML-DSA certificate listed in the public registry.

03

Go Live

Display your verified badge. Anyone can verify your agent instantly.

The Missing Layer

Why AIGP-Σ?

AI agents are autonomous. They hold keys, execute transactions, communicate with other agents — often with no human in the loop. Yet no existing standard was built to identify, authorize, or revoke them.

What exists today — and where it falls short

StandardWhat it doesWhat's missing for AI agents
X.509 / TLSServer & human identity via PKINo agent scope, no behavioral binding, no PQC
SPIFFE / SPIREShort-lived workload certificatesNo AI-specific semantics, no PQC by default
W3C DID / VCDecentralized self-sovereign identityNo behavioral fingerprint, no PQC
ISO/IEC 42001AI management system governanceFramework only — no machine-readable certificate
Hybrid PQC X.509X.509 + post-quantum algorithmsNo agent-specific extensions, still experimental

Every one of these solves part of the problem. None of them solves all of it.

What AIGP-Σ adds

AIGP-Σ is not a replacement for X.509. It is a new layer built above it — the same way TLS is a layer above TCP.

Post-Quantum Cryptography by Default

ML-DSA (NIST FIPS 204) for digital signatures. ML-KEM (NIST FIPS 203) planned for agent-to-agent key encapsulation. Both NIST standards finalized August 2024. Classic RSA and ECDSA are broken by a sufficiently powerful quantum computer. AIGP-Σ is not.

Agent-Specific Semantic Extensions

Permission scope, EU AI Act risk classification, and behavioral fingerprint — embedded directly in the certificate. Not in a separate config file. Not in a database. In the certificate itself, cryptographically bound.

Built-in Kill Switch (HALT Protocol)

Every AIGP-Σ certificate contains a revocation endpoint. Four escalating HALT levels: Soft → Hard (k-of-n) → Fleet → Emergency. Any agent, any deployment, revoked instantly.

W3C DID Compatibility

Each certificate maps to a Decentralized Identifier (DID), enabling interoperability with Google A2A Protocol and emerging agent identity ecosystems.

Why it matters now

The EU AI Act is in force. High-risk AI systems require documented accountability chains. Autonomous agents need verifiable, auditable identities — not just API keys and JWTs. AIGP-Σ is the cryptographic infrastructure for that world.

Open Research

Verifiable priority — timestamped on Zenodo

AIGP-Σ was first publicly disclosed on Zenodo with a permanent DOI, establishing an open-access, timestamped research record.

DOI: 10.5281/zenodo.19384051

Ready to certify your AI?

Join the first generation of certified AI agents. Get your AIGP-Σ certificate in minutes.

SSL for AI Agents

Per agent, per year — just like SSL certificates. Bitcoin anchoring included.

Free

Free

For individual developers and solo agents. Start building trust immediately.

  • Agent identity certificate
  • Public registry badge
  • 1 year TTL
Get Started Free
Coming Soon

Business

TBD/year

For organizations, fleets, and enterprise. Verified identity for all your agents.

  • Agent identity certificate
  • Organization identity verified
  • Unlimited agents under one organization
  • Public registry badge — Verified
Join Waiting List

Frequently Asked Questions

Everything you need to know about AI agent certificates.

What is an AI agent certificate?
An AI agent certificate is a cryptographic document that proves an AI agent's identity, authorized scope, and model integrity. Think of it as an SSL/TLS certificate but for AI agents — it lets anyone verify who the agent is and what it's allowed to do.
Why does my AI agent need a certificate?
Without a certificate, there is no way to verify your AI agent's identity or what it's authorized to do. An uncertified agent is like a website without HTTPS — it cannot be trusted. Certificates also help with EU AI Act compliance, scope enforcement, and the ability to instantly revoke compromised agents.
How do I certify my AI agent?
Sign up at aigpsigma.com, register your agent's identity and authorized scopes, and receive your post-quantum ML-DSA certificate instantly. Your agent will be listed in the public registry with a verified badge.
Is AIGP-Σ free?
Yes! Free certificates are available for individual developers and solo agents. The Business tier (for organizations needing verified certificates for unlimited agents) is coming soon.
What AI frameworks does AIGP-Σ support?
AIGP-Σ certificates work with any AI agent framework — OpenClaw, AutoGPT, CrewAI, LangChain, Semantic Kernel, MetaGPT, SuperAGI, or any custom agent built in Python, Node.js, or any language.
What is post-quantum cryptography and why does it matter?
AIGP-Σ uses ML-DSA (FIPS 204), a post-quantum signature algorithm. This means your certificates remain secure even against future quantum computers, which could break traditional RSA and ECDSA signatures.