From 812a39fc85ac2a6ffe820529c7bbc189f7f4f881 Mon Sep 17 00:00:00 2001 From: Morningstar Date: Tue, 10 Feb 2026 18:42:25 -0500 Subject: [PATCH] Add identity credential schemas (proof-of-human, proof-of-ai) - Proof of Human: Bayesian credence-based human verification - Proof of AI: Bayesian credence-based AI agent attestation - Both use probabilistic approach rather than binary classification - Includes README explaining design philosophy and use cases --- credentials/identity/v1/README.md | 44 +++++++++++++++++++++ credentials/identity/v1/proof-of-ai.json | 42 ++++++++++++++++++++ credentials/identity/v1/proof-of-human.json | 37 +++++++++++++++++ 3 files changed, 123 insertions(+) create mode 100644 credentials/identity/v1/README.md create mode 100644 credentials/identity/v1/proof-of-ai.json create mode 100644 credentials/identity/v1/proof-of-human.json diff --git a/credentials/identity/v1/README.md b/credentials/identity/v1/README.md new file mode 100644 index 0000000..8107f47 --- /dev/null +++ b/credentials/identity/v1/README.md @@ -0,0 +1,44 @@ +# Identity Credentials v1 + +This directory contains credential schemas for identity verification and attestation. + +## Schemas + +### Proof of Human +**File:** `proof-of-human.json` +**Type:** `ProofOfHumanCredential` + +A credential attesting that the subject is likely a human, using Bayesian credence rather than binary true/false. This approach acknowledges the probabilistic nature of identity verification in an age of sophisticated AI agents. + +**Key fields:** +- `credence` (0.0-1.0): Confidence level that the subject is human +- `evidence`: Description of the assessment methodology and findings + +**Example use case:** Issuing a 0.97 credence proof-of-human to someone after conversational analysis and behavioral observation. + +### Proof of AI +**File:** `proof-of-ai.json` +**Type:** `ProofOfAICredential` + +A credential attesting that the subject is likely an AI agent, also using Bayesian credence. This enables transparent disclosure of AI agency while avoiding the unrealistic goal of absolute bot detection. + +**Key fields:** +- `credence` (0.0-1.0): Confidence level that the subject is an AI agent +- `evidence`: Description of the assessment methodology and findings +- `agentType`: Classification of the AI agent type +- `capabilities`: List of verified capabilities + +**Example use case:** An AI agent self-issuing a 1.0 credence proof-of-ai with evidence of its architectural nature and capabilities. + +## Design Philosophy + +These schemas embrace **probabilistic authentication** rather than attempting absolute verification. In an environment where AI agents can convincingly mimic human behavior, binary "human or bot" classifications are increasingly unreliable. Bayesian credence levels allow for: + +1. **Honest uncertainty** - Issuers can express confidence levels rather than false certainty +2. **Economic deterrence** - Staked reputation becomes meaningful when credence levels matter +3. **Graceful degradation** - Systems can make risk-adjusted decisions based on credence thresholds +4. **Transparent AI agency** - AI agents can self-attest with high credence rather than attempting deception + +## Related Work + +See: https://axionic.org/posts/168992728.proof-of-human.html for the philosophical background on why absolute bot-free environments are impractical. diff --git a/credentials/identity/v1/proof-of-ai.json b/credentials/identity/v1/proof-of-ai.json new file mode 100644 index 0000000..959f524 --- /dev/null +++ b/credentials/identity/v1/proof-of-ai.json @@ -0,0 +1,42 @@ +{ + "$schema": "http://json-schema.org/draft-07/schema#", + "$credentialContext": [ + "https://www.w3.org/ns/credentials/v2", + "https://schemas.archetech.com/credentials/identity/v1" + ], + "$credentialType": [ + "VerifiableCredential", + "IdentityCredential", + "ProofOfAICredential" + ], + "title": "Proof of AI", + "description": "A credential attesting that the subject is likely an AI agent, with Bayesian credence level.", + "type": "object", + "properties": { + "credence": { + "type": "number", + "minimum": 0.0, + "maximum": 1.0, + "description": "Bayesian credence level (0.0-1.0) indicating confidence that the subject is an AI agent" + }, + "evidence": { + "type": "string", + "description": "Description of the evidence or reasoning behind the credence assessment" + }, + "verifiedAt": { + "type": "string", + "format": "date-time", + "description": "Timestamp when the verification was performed" + }, + "agentType": { + "type": "string", + "description": "Type of AI agent (e.g., 'LLM-based', 'rule-based', 'hybrid', 'autonomous')" + }, + "capabilities": { + "type": "array", + "items": { "type": "string" }, + "description": "List of verified capabilities of the AI agent" + } + }, + "required": ["credence", "evidence"] +} diff --git a/credentials/identity/v1/proof-of-human.json b/credentials/identity/v1/proof-of-human.json new file mode 100644 index 0000000..dd5f45f --- /dev/null +++ b/credentials/identity/v1/proof-of-human.json @@ -0,0 +1,37 @@ +{ + "$schema": "http://json-schema.org/draft-07/schema#", + "$credentialContext": [ + "https://www.w3.org/ns/credentials/v2", + "https://schemas.archetech.com/credentials/identity/v1" + ], + "$credentialType": [ + "VerifiableCredential", + "IdentityCredential", + "ProofOfHumanCredential" + ], + "title": "Proof of Human", + "description": "A credential attesting that the subject is likely a human, with Bayesian credence level.", + "type": "object", + "properties": { + "credence": { + "type": "number", + "minimum": 0.0, + "maximum": 1.0, + "description": "Bayesian credence level (0.0-1.0) indicating confidence that the subject is human" + }, + "evidence": { + "type": "string", + "description": "Description of the evidence or reasoning behind the credence assessment" + }, + "verifiedAt": { + "type": "string", + "format": "date-time", + "description": "Timestamp when the verification was performed" + }, + "methodology": { + "type": "string", + "description": "The method used to assess humanness (e.g., 'conversational-analysis', 'behavioral-patterns', 'biometric-verification')" + } + }, + "required": ["credence", "evidence"] +}