On 30 August 2023, the
US Copyright Office (USCO) issued a notice of inquiry (NOI)
on the Federal Register, seeking factual information and views on several
issues related to the intersection of copyright law and recent advances in
generative Artificial Intelligence (AI), including the use of copyrighted works
to train AI models, the appropriate level of transparency and disclosure with respect
to the use of the copyrighted works, the legal status of the AI generated
outputs and the appropriate treatment of AI generated outputs that mimic
personal attributes of human artists. The deadline to respond is 18 October
2023.
The USCO will use this
information to analyse the current state of the law, identify unresolved issues
and to advise the Congress.
This NOI comes after the
USCO cancelled the AI-generated artwork portions of the copyright registration
submitted by author Kristina Kashtanova in the case of Zarya of the Dawn
(see the IPKat here) and
issued (in March 2023) the “Copyright registration guide:
works containing material generated by artificial intelligence”
(as reported by the IPKat here). The USCO underlined again the importance of the “human authorship”
requirement and stated that applicants must bear the duty to disclose the
inclusion of AI-generated content in a work submitted for registration and to
provide a brief explanation of the human author’s contribution to the work.
It is worth noting
that, more recently, the principles above have been applied by the USCO and by the
U.S. District Court for the District of Columbia.
The work “Théâtre D’opéra Spatial” by Jason Allen is a two-dimensional AI-generated work.
Midjourney The Work
On 21 September 2022, Mr Jason Allen applied for copyright registration before the USCO naming himself as the author. The work was generated by the AI system Midjourney, but Mr. Allen (the applicant) did not disclose it in the application. Since the work has gathered national attention for being the first AI-generated image to win the 2022 Colorado State Fair’s annual fine art competition, the examiner asked for additional information on the use of Midjourney. The applicant argued that he had contributed significantly to the creation of this image. His “creative inputs” included about 624 text prompts and revisions of text prompts and the use of a photoshop software to remove flaws and to create new visual content. The applicant argued that he had also upscaled the image using an AI tool. Nevertheless, the applicant declined the examiner’s request to exclude from the copyright claim the features of the work generated by Midjourney. On 13 December 2022 (first refusal), the office refused the registration.
On 24
January 2023 (first request for reconsideration), the applicant requested the
USCO to reconsider its initial refusal by arguing that the examiner had misapplied
the human authorship requirement. The examiner (re)evaluated the claims and
(again) stated the work could not be registered without limiting the
application to the copyrightable part of the claim. It should be noted that the
examiner accepted as copyrightable the “visual edits” made with a Photoshop
tool, while he confirmed that the features generated by Midjourney and Gigapixel
AI had to be excluded from the application.
On 12 July
2023 (second request for reconsideration), the applicant filed additional
arguments in his new request to reconsider the application. The applicant affirmed
the creativity of his input into Midjourney by “enter[ing] a series of
prompts, adjust[ing] the scene, select[ing] portions to focus on, and
dictat[ing] the tone of the image” together with the fair use doctrine,
that “would allow for registration of the work” because it “allows
for transformative uses of copyrighted material.”
By decision of 5 September 2023
the Review Board of the USCO refused to register the Copyright claim in the
work, by recalling that:
i)
“human authorship is a bedrock of copyright”;
ii)
when analysing AI-generated material, the Office must determine when a human
user can be considered the creator of AI-generated output (thus recalling the
Guidance issued in March 2023);
iii)
the analysis will be conducted on a case-by-case basis;
iv)
the applicant must disclose AI-generated content that is “more than de
minimis” (sufficient authorship);
v)
the applicant may disclose and exclude such material through a brief description
of the AI-generated content while it is not requested to disclose the AI tools
used.
The Board
acknowledged that the prompting process may involve creativity, but this does
not mean that providing text prompts to Midjourney (even if in a large number
as 624 revisions and text prompts) “actually forms the generated images”. The grounds for refusal were different from
the Thaler case because
authorship by AI was not the issue here. The USCO found that the application
failed to meet the de minimis standard and the author had failed to disclose the
AI-generated portions of the work. The AI-generated amount of content was more
than de minimis and the applicant sought to register the entire work refusing
to disclaim the portions attributable to AI.
Comment
We may recall that in Thaler, the USCO refused to register an AI-generated work since in the application the AI-system was indicated as the author. That case focused on the question of whether an AI-system can be an author. The USCO ruled that it cannot since human authorship is necessary.
By
decision of 18 August 2023 the
U.S. District Court for the District of Columbia affirmed that the Copyright
Office’s refusal to register the AI-generated image “A Recent Entrance to
Paradise” was correct due to the lack of a human authorship and thus no
copyright existed in the first place.
From a
more practical perspective, when using Midjourney or other AI-tools, applicants
seeking a copyright registration at USCO must bear in mind that they must disclose
and explain what is protectable in their work.
No comments:
All comments must be moderated by a member of the IPKat team before they appear on the blog. Comments will not be allowed if the contravene the IPKat policy that readers' comments should not be obscene or defamatory; they should not consist of ad hominem attacks on members of the blog team or other comment-posters and they should make a constructive contribution to the discussion of the post on which they purport to comment.
It is also the IPKat policy that comments should not be made completely anonymously, and users should use a consistent name or pseudonym (which should not itself be defamatory or obscene, or that of another real person), either in the "identity" field, or at the beginning of the comment. Current practice is to, however, allow a limited number of comments that contravene this policy, provided that the comment has a high degree of relevance and the comment chain does not become too difficult to follow.
Learn more here: http://ipkitten.blogspot.com/p/want-to-complain.html