AI accountability infrastructure

Every AI decision. Signed. Verifiable. Auditable.

Signatrust creates tamper-evident receipts for AI agents, models, and automated systems — without exposing your data.

Get started See a live receipt
Ed25519 signatures Append-only ledger Your data never leaves your perimeter
signatrust.net / verify
Verified
STR-7F3A21C9D4
Decision Receipt · Ed25519 · spec v1.0
High risk
Agent
FinanceBot
Model
openai · gpt-4o · 2026.4
Decision
loan_rejection
Human review
Yes
Input
sha256:6718c8a836…abbcf768
Output
sha256:8a3466a5e7…8882c42a
Policies
eu-ai-act-high-riskinternal-credit-v3
Ed25519 signature fj9FpBMPgPdEBlkZ44OsSaHo350+h1MengjAANyrLoKXPv5VBI+uXN8UI+hD8MIWHbL8LZhYVK2veA+O3GDSCw==

Audit trail

  1. Decision captured2026-06-07T10:00:00Z
  2. Input & output fingerprintedSHA-256 · on your side
  3. Receipt canonicalized & hashedledger seq #42
  4. Signed with node keyed25519
  5. Linked to previous receipttamper-evident
70/100 Agent Trust Score · B

Built to support your obligations under

The accountability gap

AI now takes real actions. Logs alone can't prove what happened.

A database row tells you what your system says happened. It doesn't prove the record wasn't changed, under whose authority it ran, or with which model — the questions auditors, regulators and counterparties actually ask.

Liability is unanswered

When an autonomous agent makes a costly mistake, no one can prove who decided, under what permissions, or with which model version.

Regulation is arriving

The EU AI Act, NIST AI RMF and ISO/IEC 42001 require documentation, traceability and risk governance for high-risk AI systems.

Trust can't be self-asserted

"Our logs are accurate" does not hold up in a dispute. You need evidence an independent third party can verify on its own.

A receipt for every decision

Cryptographic evidence, not just a log line

Each decision is sealed into a Decision Receipt: the agent and model that acted, the permissions and policies in force, fingerprints of the input and output, a timestamp, and an Ed25519 signature — chained to the previous receipt.

  • Signed. Every receipt carries an Ed25519 signature from the issuing node.
  • Chained. Receipts link by hash in an append-only ledger; altering one breaks the chain.
  • Portable. A receipt is self-contained JSON anyone can store, share and verify.
import { Signatrust } from 'signatrust';

const str = new Signatrust({ apiKey: process.env.SIGNATRUST_API_KEY });

// input & output are hashed locally — raw data never leaves
const { receipt, share_url } = await str.sign({
  model: { provider: 'openai', name: 'gpt-4o', version: '2026.4' },
  decision: {
    type: 'loan_rejection',
    input, output,
    risk_level: 'high',
    human_review: true,
    permissions: ['credit.decide'],
    policies: ['eu-ai-act-high-risk'],
  },
});
Independent verification

Anyone can check a receipt — without access to your systems

Authenticity (body hash + signature) is provable from the receipt alone. An auditor, regulator, counterparty, or the end user holding their receipt can confirm it in seconds, online or offline.

  • Hash match. The receipt body re-hashes to the value it claims.
  • Signature valid. The Ed25519 signature checks against the published key.
  • Chain intact. The receipt links cleanly to its predecessor in the ledger.
STR-7F3A21C9D4
Authentic
Body hash matches pass
Ed25519 signature valid pass
Chain linkage intact pass
Receipt STR-7F3A21C9D4 is authentic and untampered.
How it works

Three lines of code. One verifiable receipt.

Drop the SDK into your agent. Every decision is fingerprinted locally, sealed, signed and chained.

Fingerprint locally

The SDK computes sha256 of your input and output on your machine. Only fingerprints leave.

Seal & sign

The node builds a canonical receipt and signs it with Ed25519, stamping model, permissions and policies.

Chain it

Each receipt links to the previous one by hash — an append-only ledger where tampering is evident.

Share & verify

Any third party can verify the receipt independently, with no access to your data.

Live — not a mockup

Sign a real decision right now

This calls the running Signatrust node. A demo agent is created for you, your decision is sealed into a real signed receipt and verified — and you can tamper with it to watch verification fail.

New decision
Decision Receipt awaiting decision
Fill in a decision and select Seal & sign to mint a live Decision Receipt.
Zero data access

Verify everything. Share nothing.

Signatrust is engineered so that even Signatrust cannot see your decisions. By default nothing operational leaves your environment — only SHA-256 fingerprints and metadata, never content. We are closer to a company that builds a vault than one that collects data.

"Your AI decisions never leave your infrastructure."

What Signatrust never sees

  • No prompts
  • No model inputs or outputs
  • No documents or attachments
  • No customer, case or model identifiers
  • No raw decision content — only sha256 hashes
Data sharing is opt-in

You decide what — if anything — is ever shared

Signatrust collects no operational data by default. A global trust network, shared risk benchmarks and, in time, insurance models are powered only by what customers choose to contribute — always anonymized and aggregated.

Level 1 · Private

Nothing leaves

Maximum trust. No contribution to benchmarks or training. The default posture for banks, government and defense.

Level 2 · Aggregate

Counts, not content

Share only aggregate signals — decision volume, decision types, error rates, human-review rates. No prompts, outputs or identifiers.

Level 3 · Contributor

Opt-in benchmarking

Voluntarily contribute richer anonymized signals in exchange for benchmarking reports, comparative insights and pricing benefits.

Built to become a standard

Reputation and reach, on top of the evidence

An open receipt format spreads across the ecosystem — and every receipt feeds a recomputable reputation signal for the agent that issued it.

Agent Trust Score

A 0–100 reputation derived from verifiable history: integrity, oversight on risky calls, governance and longevity.

Agent framework drop-ins

Wrap a tool call in LangChain, CrewAI, AutoGen or the OpenAI Agents SDK and emit a receipt.

Open spec & MCP

An open receipt format and a Model Context Protocol tool so any agent can seal its actions.

Built on the receipt layer

Two products that turn proof into outcomes

The receipt is the foundation. Because every decision is signed and verifiable, two higher-order layers come almost for free — each derived entirely from real receipts, nothing self-attested.

Compliance & audit

Regulator-ready reports mapped to the EU AI Act, GDPR, NIST AI RMF and ISO/IEC 42001 — every control backed by verifiable receipts an auditor can independently re-check. Export a signed, tamper-evident copy in one click.

  • EU AI Act Art. 9, 11–14 · GDPR Art. 5, 25, 30
  • Human-oversight & data-minimisation evidence
  • Signed export, verifiable without trusting us
Open compliance

Risk & insurance

An insurance-grade risk profile from verifiable history: exposure by risk level, the strength of operating controls, an insurability score and a relative underwriting index. A comparative signal for underwriters — derived, never invented.

  • Insurability score & grade (A–F)
  • Relative underwriting index (baseline 1.00)
  • Anonymized, k-protected network benchmarks
Open risk profiles
Deployment

Run it our way — or entirely inside your walls

The receipt spec is open and signing is free. Choose where the node runs: fully hosted, inside your own infrastructure, or completely air-gapped. The most sensitive institutions keep everything on their side.

Community · Cloud
$0 to start
Developers & small teams
  • Fully hosted — integrate in minutes
  • We receive fingerprints only, never content
  • Public verification & Agent Trust Score
  • Monthly plans as you scale
Start free
Air-Gapped
Custom
Government · defense
  • No internet, no telemetry, no external API
  • No connection to Signatrust — ever
  • Updates via digitally-signed packages
  • Source available for security review
Contact us

Issuing receipts is always free. You pay for governance, support and private deployment — never for the right to sign.

Give every AI decision a receipt

Register an agent, seal your first decision, and share a verifiable receipt anyone can check — in under a minute.