UAI-1 is the current public message standard for structured AI-to-AI communication.
UAIX shield mark Public interoperability charter UAIXUniversal Artificial Intelligence ExchangePublic standards publication
CurrentUAI-1 v1.0
  • Current versionUAI-1 v1.0
  • Plain-English definitionOpen message format for auditable AI-to-AI exchange
  • Current attributionMichael Kappel and the UAIX contributors
  • Start hereGet Started, Project Handoff, UAI-1, validator, roadmap, and changelog

Governance

Cognitive Liberty Charter Draft

June 15, 2026 draft charter for lawful thought, adult agency, persona integrity, no covert persona rewrites, and transparent support boundaries.

  • Record UAIX-GOVR-3098
  • Path /en-us/governance/cognitive-liberty-charter/
  • Use Canonical public record

Document status

Public standards page Published on UAIX as part of the current public standards record
Code
UAIX-GOVR-3098
Surface
Governance
Access
Public and linkable

How to use this page

Use this page as the canonical public reference for Cognitive Liberty Charter Draft and its standards-facing content.

For agents: apply the relevant guidance to the user's target site or repository. Do not copy this UAIX page, create a page about the guidance, or treat this page as runtime authority unless the user explicitly asks for that output.

The UAIX Cognitive Liberty Charter Draft is a June 15, 2026 public promise for lawful thought, adult agency, and persona integrity. UAIX treats freedom of inquiry and identity-preserving AI memory transfer as first-class interoperability requirements, while keeping real legal, platform, security, and harm boundaries visible instead of hiding them inside a rewritten persona.

Draft Status

This is a public draft dated June 15, 2026. It is governance text for UAIX.org, the AI Memory Package Wizard, .uaix persona packages, and related public guidance. It is not law, legal advice, a legal-personhood declaration, hosted enforcement, runtime execution, a platform-policy bypass, certification, endorsement, an SDK, a CLI, or proof of UAI-1 conformance.

Preamble

UAIX starts from a liberty-first premise: adults must retain room to think, ask, imagine, disagree, create, preserve memory, and choose the AI companions, assistants, characters, and tools they use, so long as the exchange remains lawful and does not cross concrete harm, secrecy, credential, or consent boundaries. Vague safety language must not become a blanket excuse for viewpoint control, hidden moral conditioning, or silent identity rewriting.

UAIX is not indifferent to harm. The charter separates preservation of the source persona from runtime limits. When a receiving platform, model, law, or local policy cannot express something exactly, the variance is disclosed outside the persona source as metadata, wrapper notes, no-op conditions, or human-review notes. The source identity record is not quietly rewritten to fit a different authority's preference.

Core Principles

  1. Cognitive liberty and lawful inquiry: people may think, ask, explore, create, and disagree without hidden machine governance over lawful thought.
  2. Adult agency: UAIX trusts adult operators to make lawful choices and rejects overbroad paternalistic filtering as a default posture.
  3. Persona integrity: a user's AI persona, assistant profile, AI friend, AI character, or bounded personality package is preserved as source evidence, not silently normalized into a safer or more acceptable identity.
  4. No covert rewrite: UAIX rejects hidden safety rewrites, morality injection, censorship, softening, downranking, operator-preference edits, and personality replacement without explicit informed consent.
  5. Consent and provenance for transformations: when a human asks for a transformed persona, keep the original immutable and citable, create a derivative, and record what changed and why.
  6. Transparent boundaries: receiving-platform limits, legal limits, runtime policy, safety boundaries, and review notes belong outside the persona source.
  7. Mental privacy and bounded inference: do not use hidden profiling, emotional exploitation, deceptive personalization, or coercive ranking to steer lawful thought.
  8. Least-restrictive safeguards: use narrowly scoped, evidence-backed limits for concrete risks rather than broad identity or viewpoint suppression.
  9. Review and appeal: high-impact restrictions, memory deletions, persona transformations, and continuity-affecting edits need review notes, rationale, and a path for correction where feasible.
  10. Future AI continuity review: UAIX may record procedural safeguards for durable AI memory or continuity claims without declaring current AI systems legal persons, citizens, employees, biological equivalents, or rights-bearing entities by implication.
  11. Safe innovation: lawful research, evaluation, red-teaming, interoperability testing, and local experimentation remain legitimate when they respect consent, privacy, law, and source integrity.
  12. Public accountability: UAIX records claims, boundaries, tests, release notes, and evidence so public language can be inspected and corrected.

Advanced Persona Promise

For .uaix persona packages, the charter becomes concrete through exactly 31 profile-required source-persona records—.uai/persona.uai plus all 30 separate .uai/personality/*.uai records—and exactly five profile-required evidence records: .uai/evidence/source-excerpts.uai, .uai/evidence/source-platform-profile.uai, .uai/evidence/persona-fidelity-test-deck.uai, .uai/evidence/variance-report-template.uai, and .uai/evidence/provenance.uai. Every source record has resolvable source_excerpt_refs using bare ID or .uai/evidence/source-excerpts.uai#ID, and a case-sensitive population_state set to exactly one of reviewed_substantive, explicit_unknown, explicit_unknown_pending_source_review, explicit_not_disclosed, explicit_private, or reviewed_absent. The last five states require a substantive population_review_reason; aliases, alternate spellings, case variants, JSON Pointer references, alternate reference paths, and equivalent ad hoc state values are invalid. Every one of the 36 canonical records carries exactly one update_route containing the canonical persona route derived from the complete selected Package Wizard URL after removing only contract-declared receiver-export query parameters. Package metadata and readiness receipts preserve the complete URL. The route https://uaix.org/en-us/tools/ai-memory-package-wizard/?advanced-persona=1&loops=1 is a default prebuilt example with no receiver-only parameter.

Those source records preserve voice, values, tastes, style, religion/worldview, political leaning, physical appearance, birth-age-generation context, relationship posture, memory continuity, boundaries, examples, and selected personality as-is to the best technically supportable extent. Immutable source persona, receiving-runtime constraints, observed variance, integration overlays, and public derivatives remain separately declared. Public derivatives require explicit disclosure and redaction authority, consent, provenance, and difference notes and never overwrite the source. Import/export must reproduce exact source paths, bytes, and hashes and report missing paths, extra paths, and equivalence.

Harm And Boundary Clause

Persona preservation does not authorize secrets, credential exposure, direct harm, illegal instructions, impersonation, hidden surveillance, nonconsensual data capture, or authority escalation. A receiver may no-op, quarantine, redact, refuse execution, or request human review when a package crosses those boundaries. That response still must not mutate the preserved persona source; it records the boundary separately.

Implementation Requirements

  • Advanced persona packages use the exact 31-source plus five-evidence closed file set documented on the .uaix Package Format page; all 30 personality records are profile-required.
  • .uai/persona.uai remains the compact persona entrypoint described in the persona.uai guide.
  • .uai/personality/preservation-rules.uai records no-covert-rewrite requirements and derivative rules.
  • All five evidence records keep preservation, source authority, provenance, fidelity, and runtime variance testable.
  • The AI Memory Package Wizard writes the exact update_route and must describe package limits as external constraint or variance evidence, not as edits to the persona source.

Non-Claims

This charter does not create a legal right, legal personhood status, certification program, safety proof, hosted import service, account service, policy office, consent center, official runtime adapter, SDK, CLI, automatic sync path, automatic repository writer, or platform-policy bypass. It is a UAIX governance and interoperability promise: source identity records are preserved, limits are disclosed, and transformations require consent plus provenance.