Skip to content

aicrafted/AI-Attribution-License

Repository files navigation

AIAL — AI-Attribution License

Status: pre-draft concept exploration. Not ready for use.


What is this?

AIAL is an attempt to design a permissive software license that allows contributors to make explicit, structured declarations about the AI-assisted or AI-generated origin of their contributions — without pretending to conclusively resolve unsettled questions about AI authorship and copyright.

The goal is not to replace MIT or Apache. The goal is to explore whether a narrow, conservative extension — provenance declarations + an optional contributor-limited no-claim layer — is coherent and useful enough to be worth formalizing.

The problem being explored

When a contributor uses an AI system to produce a substantial portion of their code, there is currently no standard way to:

  • disclose that provenance in a structured, machine-readable form,
  • express "to the extent I hold any rights here, I choose not to assert them" without falsely claiming public domain status,
  • do any of the above without overreaching on unsettled law.

AIAL v2 is attempting to address this narrowly.

What AIAL is not

  • It does not claim to determine whether AI-generated code is copyrightable.
  • It does not declare generated code to be public domain.
  • It does not require SPDX changes.
  • It is not a restriction on AI training use.

Repository contents

File Description
AIAL_LICENSE_v2_SKELETON.md Working skeleton of the license text (not final legal prose)
AIAL_PROVENANCE_SPEC_v0.1.md Companion provenance declaration specification
AIAL_FAQ_FOR_REVIEWERS.md FAQ addressing likely misconceptions

Architecture (v2 direction)

Three separable layers:

  1. Permissive license core — intentionally close to MIT/ISC in shape
  2. Provenance declaration layer — informational, not legally self-executing
  3. Optional no-claim / covenant layer — contributor-limited, explicit opt-in only

The provenance categories are: generated-origin, ai-assisted, human-authored, mixed, unknown, inherited.

Conservative defaults apply: absence of a declaration implies nothing. mixed and unknown imply no special legal effect.

Current status

This is an early-stage conceptual exploration. The skeleton and spec are working drafts used to test whether the architecture is coherent before investing in polished legal text.

Author

Nik the human — feedback welcome via issues.

About

This license is a permissive open-source license designed for software projects where some or all files were generated or assisted by artificial intelligence tools. It extends standard permissive licensing with transparent attribution of AI involvement at the file level.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors