Skip to content

ĀML Core 1.4.0-rc.2 — Hardened Federated Trust & Adoption Preview

Pre-release
Pre-release

Choose a tag to compare

View all tags
@github-actions github-actions released this 08 Sep 05:51
v1.4.0-rc.2
675b85a

ĀML Core v1.4.0-rc.2 — Hardened Federated Trust & Adoption Preview

v1.4.0-rc.2 refreshes the public post-v1.3 architecture snapshot with the latest security, documentation, trust-root, ecosystem, and adoption hardening from main.

Version boundary: the stable package/CLI/capability version remains v1.3.0. This is a GitHub prerelease snapshot of the broader v1.4 candidate architecture, not a claim that stable v1.4 has been published to a package registry.

What RC2 adds over RC1

  • hardened SECURITY.md covering compiler/parser, policy, cryptography, federation, capability authority, HTTP, browser, and official-brand trust surfaces
  • complete JavaScript API documentation matching the actual exported reference implementation
  • current roadmap separating stable v1.3 from v1.4 release-candidate work
  • corrected public conformance-version drift
  • expanded ecosystem map across adoption, federation, trust, standards, security, browser, and official authorization
  • Why AML Now adoption brief for developers, AI teams, researchers, enterprises, journalists, and standards participants
  • explicit production ĀRU trust-root/public-key discoverability
  • clarified changelog labeling so prerelease work cannot be mistaken for stable v1.4

Architecture carried forward from RC1

Mainstream adoption

  • AI Interface Firewall™
  • View Meaning™ browser inspector
  • privacy-minimal View Meaning browser-extension prototype
  • React-compatible accountable UI adapter
  • plain-JavaScript, React, and Next.js starter kits
  • Meaning Gate™ GitHub Action
  • Verify Official ĀML GitHub Action
  • dependency-free AML HTTP service with OpenAPI contract
  • 30-minute enterprise pilot kit

Trust continuity

  • expiring and revocable consent ledgers
  • policy consensus with explicit dissent
  • semantic risk scoring
  • multi-policy matrices
  • signed audit checkpoints
  • attention-ledger integrity verification
  • execution provenance graphs
  • Merkle-batched receipt inclusion proofs

Cross-system interoperability

  • runtime capability negotiation
  • portable policy passports
  • content-addressed bundles
  • selective-disclosure commitments
  • aml-wire/1
  • federated AML exchange
  • causal execution graphs
  • canonical JSON serialization
  • golden protocol vectors
  • replay resistance

Trust fabric

  • delegated trust chains
  • M-of-N threshold authorization
  • append-only transparency logs
  • bounded signed capability tokens
  • revocation registries
  • Proof-Carrying Interface™ manifests
  • machine-verifiable conformance claims
  • layered compatibility from Core through Governed

Official AML verification

  • machine-readable mark and authorization registries
  • signed aml-brand-authorization/1 credentials
  • canonical ĀRU trust-root registry
  • browser, CLI, HTTP, and GitHub Action verification
  • first production ĀRU Ed25519 public verification key/fingerprint
  • production private signing material intentionally kept outside GitHub

A self-signed credential can be cryptographically valid while still not being an official ĀRU authorization. Official verification additionally requires an active, non-revoked signer fingerprint in the canonical ĀRU trust-root registry.

Standards surface

  • RFCs 0001–0011
  • protocol discovery
  • versioned registries
  • JSON Schemas
  • conformance levels and fixtures
  • canonical vectors
  • independent second-runtime challenge
  • explicit governance and standardization path

Try AML now

Verify the repository

git clone https://github.com/aruintelligence/aml-core.git
cd aml-core
npm install --ignore-scripts
npm run check:links
npm test
npm run benchmark

The AML Conformance workflow independently checks the public conformance surface.

Open-source and brand boundary

Covered software remains MIT licensed. Official ĀML™ / ĀRU™ branding, logos, compatibility branding, certification-style identity, endorsement, OEM/co-branding, and related reserved brand rights are separate from the software license.

Technical conformance does not by itself imply official authorization, endorsement, partnership, certification, or trademark rights.

Evidence boundary

ĀML improves software-level inspectability, policy control, reproducibility, interoperability, and cryptographic integrity. It does not establish that current attention/restoration values objectively measure cognition or wellbeing, does not make an AI's declared intent truthful, and does not replace WCAG conformance or assistive-technology testing.

Cryptographic signatures prove integrity and key possession. They do not prove moral correctness or institutional trust by themselves.

Created by Daniel Jacob Read IV and stewarded by ĀRU Intelligence Inc.™.

Assets 2
Loading