sourceactiveEvaluation and media provenance Assets and reference collections

AIと著作権に関するチェックリスト＆ガイダンス (Checklist and Guidance on AI and Copyright, Agency for Cultural Affairs, 31 July 2024)

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 1 · updated 2026-10-18 · src_01M569MMB4SWVC25XAVR518V0Y

Evidence: review revision unknown · inspection source_inspected · execution not_run · review unreviewed · origin research_agent

Source: https://www.bunka.go.jp/seisaku/chosakuken/pdf/94097701_01.pdf

Page count and the rights-holder guidance (Part 2) were not read in full; the stakeholder structure, Part 1 sections 1-1-1 to 1-1-6 and the disclaimer were read.

Structured fields

{
  "revision": {
    "ref": "令和6年７月３１日 (31 July 2024); HTTP Last-Modified of the PDF observed 2024-08-21",
    "pinned": true,
    "source_url": "https://www.bunka.go.jp/seisaku/chosakuken/pdf/94097701_01.pdf"
  },
  "checked_at": "2026-10-17T00:00:00.000Z",
  "usefulness": "Agency for Cultural Affairs' practical explainer organised by stakeholder (AI developer, provider, business user, non-business user, and a separate guidance part for rights holders), with concrete risk-reduction steps such as checking the non-enjoyment-purpose requirement, respecting robots.txt-type access restrictions and not training on known pirate sites. A step list agents can map onto a data-collection pipeline.",
  "source_type": "documentation",
  "availability": "available",
  "source_owner": "文化庁著作権課 (Copyright Division, Agency for Cultural Affairs)",
  "next_check_at": "2026-11-16T00:00:00.000Z",
  "known_limitations": [
    "It says that following the steps is neither required to avoid nor sufficient to rule out legal responsibility.",
    "Builds on the 考え方 (March 2024), the Cabinet Office interim summary and the METI/MIC AI Business Guidelines v1.0; later versions of those were not checked.",
    "Japanese only; no English version of the checklist was found on the Agency landing page.",
    "Not legal advice."
  ],
  "upstream_updated_at": "2024-08-21T05:11:04.000Z"
}

Evidence claims

[sourced_fact] The document is by the Copyright Division of the Agency for Cultural Affairs, dated 31 July 2024, and sets out measures by stakeholder role; it builds on the 考え方, the Cabinet Office interim summary and the METI/MIC AI Business Guidelines. · locator: cover and はじめに p.1 source
[sourced_fact] It states that the listed risk-reduction measures neither create legal liability if skipped nor guarantee freedom from liability if all are followed. · locator: はじめに, 資料の構成について (p.5) source
[sourced_fact] AI developers are told to check that collection of training data meets the non-enjoyment-purpose requirement of Article 30-4, listing three cases where enjoyment purpose may coexist (intentional over-training to output the work, RAG databases built to output the work, and small-scale additional training on one creator's works such as LoRA). · locator: 第1部 1-1-1 (p.7) source
[sourced_fact] Copying a paid-for database work offered as an AI training dataset without permission may fall within the Article 30-4 proviso so that the article does not apply. · locator: 第1部 1-1-2 (p.8) source
[sourced_fact] Technical measures such as ID/password access limits or robots.txt, together with past sales, can support an inference that a database work is planned to be offered as paid training data; circumventing them may take the collection outside Article 30-4. · locator: 第1部 1-1-3 (p.8) source
[sourced_fact] Developers are told not to knowingly collect training data from pirate sites, since responsibility for resulting infringement may attach to the developer. · locator: 第1部 1-1-5 (p.9) source
[sourced_fact] Roles overlap: an individual who trains and publishes a LoRA and also uses it should read the developer, provider and non-business-user sections. · locator: 資料の構成について (p.5), example 2 source
[sourced_fact] MEXT website rules, which the Agency site points to for reuse of its content, require a source credit (出典) when using content. · locator: 文部科学省ウェブサイト利用規約 source
[proposal] A data-collection agent can turn sections 1-1-1 to 1-1-6 into pre-flight checks (purpose log, paid-dataset check, robots.txt/ID-password respect, pirate-source denylist, license fallback). · locator: sections 1-1-1 to 1-1-6 source
[upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.

Rights

component	license	hosting	redistribution	modification	training	Evidence and notes
documentation	LicenseRef-Japan-PDL-1.0-via-MEXT-rules	permitted	permitted	permitted	unknown	https://www.bunka.go.jp/bunkacho_homepage/index.html https://www.mext.go.jp/b_menu/1351168.htm https://www.digital.go.jp/resources/open_data/public_data_license_v1.0 Same Agency-site terms as the 考え方 record: MEXT website rules (PDL 1.0, CC BY compatible), 出典 required, third-party-rights content excluded. The terms do not address AI training. L07 2026-10-17: PDL 1.0 text read at digital.go.jp (rules set 2024-07-05): content may be reproduced, publicly transmitted, translated and adapted, including commercially, with a source credit (出典); CC BY 4.0 compatible; third-party-rights content excluded. Normalized to permitted (attribution), as OGL v3.0 and CC BY are elsewhere in the collection.

Linked records

Machine-readable: JSON · Markdown · revisions · changes