Copyright Protection For AI-Generated Cross-Lingual Literary Translations.
1. Understanding AI-Generated Cross-Lingual Translations
A cross-lingual literary translation is the conversion of a text from one language into another while maintaining the meaning, style, and literary elements. Traditionally, human translators have been the authors of these works and thus entitled to copyright protection.
When AI is used to produce translations (for example, using machine translation models), the legal questions arise:
Can AI-generated works be copyrighted?
Who owns the copyright: the user, the developer, or no one at all?
How is originality judged in AI-generated translations?
Key legal principles involved:
Originality: Copyright requires human creativity. Purely automated outputs may lack the "human author" element.
Fixation: The work must be fixed in a tangible medium.
Derivative works: Translations are often considered derivative works, and copyright may extend only with permission from the original author.
2. Key Case Laws on AI-Generated Works and Translation
Since AI-generated translations are a recent phenomenon, courts have addressed related issues through derivative works, machine-assisted creation, and originality. Here are detailed cases:
Case 1: Naruto v. Slater (Monkey Selfie Case, 2018, US)
Facts: A macaque took selfies with a photographer’s camera. PETA argued the monkey should have copyright over the images.
Ruling: The court held that non-human authors cannot hold copyright.
Implication: AI, being non-human, is similarly unlikely to be recognized as an author under current US copyright law.
Relevance to AI translation: Any translation generated purely by AI without human creative input is likely not copyrightable in the US.
Case 2: Thaler v. Commissioner of Patents (DABUS AI, Australia 2021 & UK 2022)
Facts: Thaler applied for patents for inventions generated by his AI system DABUS, claiming AI as the inventor.
Ruling: Both the UK and Australia rejected AI as a legal inventor; patents require a human inventor.
Implication for translations: AI alone cannot claim authorship, reinforcing the principle from the Naruto case. The person using AI may claim copyright only if there is sufficient human creative contribution.
Case 3: Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991)
Facts: A telephone directory publisher copied data from another directory. The court had to decide if the selection/arrangement could be copyrighted.
Ruling: Facts themselves are not copyrightable, but creative selection or arrangement is.
Relevance: In AI-assisted translation, if the human translator selects, edits, or refines AI output, this human creativity could make the translation copyrightable. Pure AI output without human selection may not qualify.
Case 4: Bridgeman Art Library v. Corel Corp., 36 F. Supp. 2d 191 (SDNY 1999)
Facts: Corel used photographs of public-domain artworks from Bridgeman without permission.
Ruling: Exact reproductions of public-domain works are not copyrightable because there is no originality.
Implication for AI translations: Literal word-for-word translations generated by AI may lack originality, similar to exact reproductions. Human input is necessary to meet originality standards.
Case 5: Narula v. State (India, 2022, AI Translation in Publishing)
(Fictional illustrative case reflecting ongoing debates in Indian law)
Facts: A publisher used an AI tool to translate a popular novel into Hindi without human revision and claimed copyright.
Ruling: The court emphasized that mere machine translation does not involve human creativity. The AI operator cannot claim copyright unless they added substantial human literary judgment.
Implication: Indian copyright law (Copyright Act, 1957) aligns with international principles: AI alone cannot hold copyright.
Case 6: Authors Guild v. Google, Inc., 804 F.3d 202 (2d Cir. 2015)
Facts: Google scanned books and created digital copies; authors claimed copyright infringement.
Ruling: Fair use applied because Google’s use was transformative.
Relevance: If AI translations are transformative (e.g., improving accessibility, cross-lingual adaptation), there may be room for fair use. However, mere literal AI translation without human creative input is unlikely transformative.
3. Principles Emerging from the Cases
AI cannot be an author. (Naruto, Thaler)
Human creativity matters. Substantial human editing or literary judgment is needed for copyright.
Derivative works require original permission. Pure machine translation may be derivative without originality. (Feist, Bridgeman)
Transformative use may allow fair use claims. (Authors Guild v. Google)
International consistency: Most jurisdictions, including US, UK, India, and Australia, emphasize human authorship as the threshold for copyright.
4. Practical Implications for Cross-Lingual AI Translations
AI-Assisted Translation: Human translators editing AI output can claim copyright.
Pure AI Translation: Typically not copyrightable; may only be protected under contract or trade secrets.
Derivative Rights: Translating copyrighted works requires permission, regardless of AI involvement.
Best Practices: Maintain records of human editing to demonstrate originality for legal protection.
Summary:
AI-generated translations exist in a legal grey area. Courts consistently emphasize human authorship as the cornerstone of copyright. To secure protection, a human must add creativity, select expressions, or transform the AI output substantially. Pure AI output is generally not copyrightable, though its use may still be licensed under other IP frameworks.
RELATED Blog
comments