◆ Semantic Web Catalogue

Problem thread

On what terms may anyone reuse it?

What licence, if any, should published data carry, and who gets to set the terms of its reuse?

issue
governance
status
reopened
steps
10, 2005–2026
threads
all problem threads

Linked data is only useful if others may reuse it, and copyright and database rights apply by default. The open-knowledge community defined what 'open' means, wrote licences and machine-readable ways to attach them, and split between waiving all rights (CC0, chosen by Wikidata) and share-alike for databases (ODbL, chosen by OpenStreetMap). Large-scale AI training has reopened the question of what reuse a licence is meant to allow.

Lineage

  1. Oct 2005
    proposes standard ★ seminal
    Open Definition

    Defines 'open' for content and data: anyone can freely access, use, modify and share it, subject at most to requirements that preserve provenance and openness. [src]

  2. 1 May 2008
    proposes standard
    ccREL

    An RDF vocabulary for describing licences and attaching them to works in web pages, so software can read the terms. [src]

  3. 2009
    proposes standard
    CC0

    A way to waive all copyright and related rights, database rights included, as far as the law allows, so data can be reused and combined without conditions. [src]

  4. 2010
    proposes concept ★ seminal
    Linked Open Data

    Linked Open Data is Linked Data released under an open licence; the first star of the five-star scheme requires an open licence before anything else. [src]

  5. Sep 2012
    proposes standard
    Open Database License

    Share-alike for databases: OpenStreetMap completes its move from a Creative Commons licence to the Open Database License, which requires adapted databases to stay open. [src1][src2]

  6. 29 Oct 2012
    settles project ★ seminal
    Wikidata

    All structured data in Wikidata is published under CC0, so anyone may reuse it for any purpose without attribution. [src]

  7. Mar 2016
    proposes concept ★ seminal
    FAIR data principles

    For research data to be reusable, data and metadata should be released with a clear and accessible data usage licence. [src]

  8. 15 Feb 2018
    proposes standard
    ODRL

    A W3C policy language for stating permissions, prohibitions and duties over content and data, more fine-grained than a licence name. [src]

  9. 25 Jun 2025
    reopens concept
    CC signals

    Creative Commons proposes signals that let dataset holders state how they want their content reused by machines, a response to AI training; partly legally binding and partly a social expectation. [src]

  10. 24 Sep 2026
    proposes writing
    W3C Workshop Report: The Future of ODRL

    A W3C workshop reaches consensus on a new Working Group for ODRL 3.0: machine-actionable policies across sectors, with formal semantics, conformance rules and a test suite. [src]

Recent signals

Dated events that show the problem moving, newest last.

  1. 1 Jul 2025

    Cloudflare blocks AI crawlers by default on new sites, turning a site's terms into an enforced gate.[src]

  2. Aug 2026

    The IETF's AI Preferences working group revises its vocabulary for stating how content may be used by AI, on track for a proposed standard.[src]

  3. 7 Sep 2026

    The ODRL Community Group settles a draft charter for a W3C 'Digital Policy Working Group' to take ODRL to version 3. ODRL[src]

Open now

Under each question, the answers people have put forward or built. The catalogue gives no odds on them.

  • Can a licence or policy say anything enforceable about use for AI training? [src1][src2]
    CC signals are described as legally binding in some cases and normative in others; a July 2026 W3C workshop agreed to start ODRL 3.0 to make policies machine-actionable.
    • A standard vocabulary states AI-use preferences, attached to content.
      The IETF AI Preferences drafts define terms such as training and search use that a site can allow or refuse.[src]
    • The law gives an opt-out its force.
      The EU's 2019 copyright directive lets rights holders reserve text and data mining in machine-readable form.[src]
    • Policies become machine-actionable, with formal semantics and tests. W3C Workshop Report: The Future of ODRL
      The 2026 ODRL workshop agreed on a new working group for ODRL 3.0.[src]
  • Should open data waive all rights or require that derived data stay open? [src1][src2]
    Wikidata chose CC0 and OpenStreetMap chose the share-alike ODbL.
    • Waive all rights, so anyone can reuse the data. CC0
      Wikidata publishes its data under CC0.[src]
    • Require that derived databases stay open. Open Database License
      OpenStreetMap licenses its data under the ODbL, which requires share-alike for derived databases.[src]

Forces

Pressures that push this problem one way or another, as their sources argue them. All forces

How to contribute