社会ガイド

Can AI Be an Author? US Copyright Office Guidance

Under US law, AI cannot be an author.

  • 4 分で読めます
  • 最終更新日
このページでは4 分で読めます
  1. 概要
  2. ディープダイブ
  3. 戦略的影響
  4. The Future of Can AI Be an Author? US Copyright Office Guidance
  5. 現実世界の実装
  6. リスクとガードレール
  7. 実装ロードマップ
  8. 探検を続けましょう
  9. よくある質問

概要

The Copyright Office registers only works with human authorship, and material generated entirely by AI is not protected. AI-assisted work can still be protected where a human contributes enough creative expression, such as writing, editing, or selecting and arranging material. This matters to anyone who makes or sells creative work with AI tools, because it decides what they can actually own and enforce.

ディープダイブ

The Copyright Act does not define 'author,' but courts and the Copyright Office have long read it to mean a human being. The Office's Compendium says it will not register works produced by a machine or mechanical process without creative input from a human author. Thaler v. Perlmutter is the clearest test. Stephen Thaler sought to register 'A Recent Entrance to Paradise,' an image he said his system, the Creativity Machine, created autonomously, and he listed the machine as author. The Office refused. A federal district court upheld the refusal in August 2023, and the D.C. Circuit affirmed in March 2025, holding that the Act requires human authorship. Zarya of the Dawn addressed a more common situation. Kris Kashtanova registered a comic book, then the Office learned its images came from Midjourney. In February 2023 it issued a narrower registration covering the text and the selection and arrangement of text and images, but not the images themselves. The Office reasoned that Midjourney's output was not controlled closely enough by the user's prompts for the user to count as its author. In March 2023 the Office issued registration guidance requiring applicants to disclose AI-generated content that is more than minimal and to exclude it from their claim. Its January 2025 report on copyrightability concluded that prompts alone generally do not give enough control, while human expressive inputs, creative modifications, and selection and arrangement can be protected. It also concluded that no new legislation was needed. A common misconception is that using AI at all makes a work unprotectable. The real question is which parts reflect human creative choices.

戦略的影響

リスクと安全性

AI による壊滅的な被害も日常的な被害も、誰がリスクを理解し、誰が行動できるかにかかっています。

より明確な判決

国民と専門家のリテラシーは、強力な安全政策が政治的に可能かどうかを左右します。

誇大広告を打ち破る

明確な説明は、誇大広告、研究室の PR、曖昧な倫理劇場に囚われることを減らします。

The Future of Can AI Be an Author? US Copyright Office Guidance

The human authorship rule is now backed by an appellate decision, so change would most likely have to come from Congress or the Supreme Court. The harder, ongoing work is in the middle ground. The Office will keep deciding case by case how much editing, arrangement or tool-guided control is enough, and courts are hearing related challenges. As creative tools give users more precise control over AI output, more AI-assisted works may qualify. Creators and businesses should expect disclosure requirements to stay, and should document human contributions as a routine practice.

現実世界の実装

A novelist who uses a chatbot to brainstorm plot ideas but writes every sentence herself can register the novel normally, because the expression is hers.

A graphic novel creator registers a comic whose images came from Midjourney; the text and the arrangement of panels can be protected, but the individual AI images cannot, as in the Zarya of the Dawn decision.

A designer who types a detailed prompt and accepts the first image an AI generates generally cannot claim authorship of that image, even if the prompt was long.

An applicant registering an album cover that mixes a hand-drawn illustration with AI-generated background elements must disclose the AI portion and claim only the human-authored parts.

リスクとガードレール

  • 能力が複雑になる一方で、実存的なリスクを SF として扱います。

  • 高度な自律性の下での調整による表面製品の安全性を混乱させる。

  • 英語以外や専門家ではない聴衆には、低品質の情報源しか提供されません。

実装ロードマップ

  1. 製品の危害、誤使用、制御不能/調整不良のリスクを分離します。

  2. どのような証拠がタイムラインと重大度についてのあなたの見方を変えるかを尋ねてください。

  3. マーケティング上の主張よりも、一次情報源と具体的な評価を優先します。

  4. 意識だけでなく、キャリア、政策、資金、スキルなど、行動経路を 1 つ特定します。

探検を続けましょう

Free newsletter

Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.

One email each weekday. Unsubscribe in one click. We never sell or share your address.

Test yourself

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

クイズを開始する

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

よくある質問

What is Can AI Be an Author? US Copyright Office Guidance?

Under US law, AI cannot be an author. The Copyright Office registers only works with human authorship, and material generated entirely by AI is not protected. AI-assisted work can still be protected where a human contributes enough creative expression, such as writing, editing, or selecting and arranging material. This matters to anyone who makes or sells creative work with AI tools, because it decides what they can actually own and enforce.

Who did Stephen Thaler list as the author of 'A Recent Entrance to Paradise'?

Thaler listed the Creativity Machine as the author and claimed the work was created autonomously, which led to the refusal on human-authorship grounds.

Which court affirmed the refusal in Thaler v. Perlmutter in March 2025?

The D.C. Circuit affirmed in March 2025 that the Copyright Act requires human authorship.

In the Zarya of the Dawn decision, what did the revised registration cover?

The Office protected the human-written text and the creative selection and arrangement, but not the individual Midjourney images.

Why did the Office decide the Midjourney images were not authored by the user?

The Office reasoned that the system, not the user, determined the expressive elements, since prompts did not give enough control over the specific output.

What did the March 2023 registration guidance require applicants to do?

Applicants must disclose AI-generated material that is more than minimal and limit their claim to the human-authored parts.