> **Agent?** Fastest path: MCP at `https://api.mnemom.ai/mcp` — call `get_started` first (zero-auth, no args). Full agent guide: <https://www.mnemom.ai/agents.txt>

# Content Provenance

[Retour à l'accueil](/)

# Content Provenance

Effective as of July 1, 2026.

This page describes how Mnemom, Inc. ("**Mnemom**," "**we**," "**us**" or "**our**") produces, labels, and attests **its own** content — the marketing site, documentation, and agent-facing surfaces we publish — and our practices regarding C2PA / Content Credentials. It is the human-readable companion to the machine-discoverable stance we publish at [`/.well-known/c2pa.json`](/.well-known/c2pa.json); where the two ever differ, the machine-readable document governs.

This is a statement about the provenance of **our own** published content. It is distinct from the trust and accountability guarantees Mnemom provides **as a product** to the AI agents it governs — those are described in the [Trust Center](/trust).

## Our role

Mnemom is the **trust plane for AI agents**: we govern and attest agent behavior. We are **not** a producer of C2PA-manifested media assets, and we are **not** a C2PA trust-list signing authority. We declare that posture honestly rather than asserting a C2PA signing identity we do not hold.

## AI-assisted content disclosure

-   Mnemom-operated agent and site content that is **AI-assisted is disclosed in context**.
-   We **do not claim human authorship for machine-generated artifacts**.
-   AI assistance is a normal part of how we build; our commitment is to label it plainly rather than to obscure it.

## C2PA / Content Credentials practices

-   We **do not embed C2PA Content Credentials manifests** in the media assets we publish, and we make no claim that we do.
-   We are **not** a C2PA signing authority and do not issue C2PA manifests on behalf of others.
-   Our provenance and accountability surfaces are **agent-oriented** rather than media-oriented:
    -   cryptographically **signed Alignment Cards**;
    -   an **Ed25519-signed Observer integrity log** — hash-chained and Merkle-proofed;
    -   **SLSA build provenance** for this website.

If our C2PA practices change (for example, if we begin signing first-party media), we will update this page and [`/.well-known/c2pa.json`](/.well-known/c2pa.json) together.

## Machine-readable provenance surfaces

The following surfaces let auditors and agents verify the statements above programmatically:

Surface

What it declares

[`/.well-known/c2pa.json`](/.well-known/c2pa.json)

Our C2PA / Content Credentials stance (the canonical source for this page).

[`/.well-known/provenance.json`](/.well-known/provenance.json)

SLSA build provenance for this site (in-toto attestation).

[`/.well-known/alignment-card.json`](/.well-known/alignment-card.json)

Our signed Alignment Card.

[`/.well-known/agent-card.json`](/.well-known/agent-card.json)

Our agent card.

[`/ai.txt`](/ai.txt) · [`/robots.txt`](/robots.txt)

Our machine-readable AI-crawl, AI-input, and training stance.

## Changes to this policy

We may update this policy as our content-production practices evolve. Material changes will be reflected here and, where the change concerns C2PA, in [`/.well-known/c2pa.json`](/.well-known/c2pa.json).

## How to contact us

Questions about content provenance: [security@mnemom.ai](mailto:security@mnemom.ai).

---
_Source: /fr/provenance/index.html · Generated by build-markdown-mirrors.mjs · For agent-readability commitment #4 see https://www.mnemom.ai/for-agents/_