METHOD FOR PRESERVING PRIMARY COPYRIGHT IN RELATION TO SECONDARY WORKS GENERATED BY ARTIFICIAL INTELLIGENCE.

Abstract

The present invention relates to a method for proving copyright in a primary work and when secondary or derivative works generated by artificial intelligence (AI) are derived from that same work. The method is useful for all creators or designers, whether individuals or professionals, as well as for private or professional organizations specializing in intellectual property, and for regulated and unregulated professions. Figure 1 provides an abstract.

Description

Title of the invention: METHOD FOR PRESERVING PRIMARY COPYRIGHT RELATED TO WITH SECONDARY WORKS GENERATED THROUGH ARTIFICIAL INTELLIGENCE.

[0001] The present invention relates to a method for proving copyright in a primary work and when secondary or derivative works generated by artificial intelligence (AI) are derived from that same work.

[0002] The process is useful for all creators or designers, whether private or professional, and for private or professional organizations specializing in IP and for regulated or unregulated professions.

[0003] In order to better understand or perceive the usefulness of the invention, it is necessary to consider the current legislative state governing the various mechanisms or processes in favour of copyright in relation to the emergence of creation assisted by artificial intelligence.

[0004] For the time being, most legislators consider that only work created by a human person deserves copyright protection.

[0005] However, in the United Kingdom, section 9.3 of the Copyright Act assigns the rights to the person who has taken the necessary steps to create the said work by means of a computer.

[0006] In France, if the creation is simply assisted by a computer, then the human person holds the copyright to his work.

[0007] On the other hand, if the creation is generated without the intervention of a human person then it cannot be assigned the qualification of "work", because in the sense of French law, the status of author is reserved for human persons.

[0008] Moreover, the works produced may sometimes bear a significant resemblance to the original creation from which the artist drew inspiration.

[0009] The use of this innovation requires advanced technical skills on the part of the creator. This may be a barrier for those who do not master the technical aspects of the technologies.

[0010] Generative artificial intelligence also raises the ethical question of data collection. This technology requires the collection of a large amount of data. It is important to remain vigilant regarding the use and collection of data.

[0011] The financial aspect can represent another obstacle to the use of this innovation. The implementation of generative AI systems can be very expensive in terms of hardware. or software. Creators working with limited budgets do not have access to this technology.

[0012] One of the other important problems related to generative artificial intelligence concerns the issue of copyright and intellectual property. When a work is generated by AI, it is difficult to determine who holds the rights to the work created via the algorithm.

[0013] To that end, let us consider how AI-generated artworks work.

[0014] In simple terms, AI-generated digital art or digital images are generated by an algorithm or machine learning process. This means that a machine has collected the data it needs (or has “learned” it) and used it to produce a new AI-generated image. All AI-generated artworks begin with the collection of vast amounts of data.

[0015] Indeed, in 2020, the High Council for Literary and Artistic Property (CSPLA) issued a report on the use of artificial intelligence in culture. The Council seems inclined to discuss this, because, according to it, "positive law must accommodate these AI-assisted creations, just as it has accommodated other technological developments in the past." From this perspective, special legislation should allow for tailored protection. (emphasis added).

[0016] "Some artificial intelligences do not hesitate to exploit images protected by copyright"

[0017] Some artificial intelligences do not hesitate to exploit copyrighted images with complete impunity.

[0018] Adobe (registered trademark), a leading name in digital media solutions, recently launched an AI-based image generation tool. It is distinguished by its data training approach. Instead of extracting data on a large scale from the web, Adobe ensures that the tool's training data is free of copyright issues by using an Adobe Stock dataset, as well as open-licensed works and public domain content whose copyright has expired. This method significantly reduces the risk of intellectual property theft.

[0019] In terms of text generation, Copyleaks offers a tool for identifying unauthorized content from IN releases. This platform aims to monitor the use of generative AI and identify copyright risks.

[0020] Through a cloud-based structure.

[0021] Therefore, in this ambiguity and as long as AGIs do not allow for greater user control over final outputs, only the reworking of the generated creation (downstream) could allow the author to leave their mark and their Personal creativity is an element of the public domain, thus restoring the exclusivity of copyright on the final work. Through these prerogatives, the author could then, in particular, oppose its exploitation by third parties and / or derive financial benefits from it. Indeed, it is essential to dedicate human intellectual and creative effort to a creation in order for it to be rewarded with copyright.

[0022] Neither the Intellectual Property Code nor current doctrine allows us to determine who owns the copyright in the event of the use of AI.

[0023] The generalization of the use of AI and its improvement requires each individual to determine on a case-by-case basis whether or not they could be considered an author and therefore benefit from rights to the work created with AI.

[0024] Pending clarification of case law, the analysis of ownership would be based on:

[0025] The general terms and conditions of the IAG used, and in particular the license and ownership clauses;

[0026] The nature and determination of the AI ​​training data used;

[0027] The framework for exploiting the generated work with regard to third-party rights;

[0028] The work carried out by humans on the generated work in order to bring their imprint of their personality to it, and thus benefit from a priori protection by copyright as a natural person author on an original work;

[0029] Documentation of the creative work carried out upstream and downstream of the generation of the work by the AI ​​in order to demonstrate all the human interventions enabling its final version (complexity of the prompt, different requests to the AI ​​on specific subjects, post-creation work, etc.).

[0030] What does the prior use of the overlock stitch remind us of?

[0031] In the United States of America, for example, the National Copyright Office has stated that only original works created by a human being are eligible for copyright protection. This position stems from case law (Feist Publications v. Rural Telephone Service Company, Inc. 499 US 340 (1991)) according to which copyright protects only the product of intellectual labor based on the creative power of the mind. Similarly, in a recent case in Australia (Acohs Pty Ltd v. Ucorp Pty Ltd), a court held that a work created using a computer could not be protected by copyright because it had not been created by a human.

[0032] In Europe, the Court of Justice of the European Union has also affirmed on several occasions, notably in a landmark judgment in the Infopaq case (C-5 / 08 Infopaq International A / S v. Danske Dagbaldes Forening), that copyright applies only to original works and that originality ranges from pair with “an intellectual creation proper to its author.” This decision is generally interpreted as meaning that an original work must be a reflection of the personality of its author, which means that human intervention is essential for a work to be protected by copyright.

[0033] The second solution, which consists of attributing authorship of the work to the programmer, is the one preferred in certain countries such as Hong Kong (SAR), India, Ireland, New Zealand, and the United Kingdom. The United Kingdom's copyright legislation best summarizes this approach. Section 9.3 of the Copyright, Designs and Patents Act stipulates that:

[0034] “In the case of a literary, dramatic, musical or artistic work created using a computer, the person who took the necessary steps to create said work shall be deemed to be its author.”

[0035] Furthermore, in Article 178, this same law specifies that “computer-generated work” means any “work created by computer under conditions excluding any human intervention.”

[0036] This provision aims to create an exception to the requirement of human authorship of the work by recognizing the amount of work invested in the design of a program capable of giving rise to a work, even if the creative spark comes from the machine itself.

[0037] Attributing authorship of a work to the person who made its creation possible through artificial intelligence seems to be the most logical approach, with the British model appearing to be the most effective.

[0038] (Source: https: / / www.wipo.int / wipo_magazine / fr / 2017 / 05 / article_0003.html)

[0039] However, it appears in the intricacies of the questioning of the authorship of literary and / or artistic works or of applied arts that current devices and processes do not resolve all of the disadvantages mentioned above.

[0040] The present invention also aims to remedy the aforementioned drawbacks by means of a method enabling proof of copyright on a primary work and when secondary or derivative works generated by artificial intelligence (AI) result from this same work, as recalled by the United Kingdom's copyright legislation, "the person who has taken the necessary steps to create said work shall be deemed to be its author".

[0041] Also, the process according to the invention makes it possible to establish the evidence and therefore the necessary provisions to be deemed the author even if secondary works generated by AI subsequently emerge.

[0042] Moreover, the works produced may sometimes bear a significant resemblance to the original creation from which the artist drew inspiration.

[0043] When creating a work, the author who uses AI to assist them or generates the secondary work based partially or completely on the primary work, the invention makes it possible to prove who owns the work, namely the creator or designer of the original work. Indeed, the author of the secondary work and the AI ​​will not be able to provide the complete document of the initial work, but only fragments or sequential parts of said work that they have accessed on the internet or in paper documents issued by the author themselves.

[0044] To avoid a conflict of interest, the inventor using AI must create the same document related to the invention as the author who does not use it. A creator or designer knows the purpose for which they create a new product.

[0045] The invention will also have the particularity of proving that it is not simply a complete copy or a reproduction improved or corrected by the author of an existing work.

[0046] The document according to the invention follows a methodology and a chronology of writing and / or completion. Assuming that the author uses part of an existing work, but that the final combination is original, even if generated by AI, the drafting of this document, its timestamping, and its deposit on the blockchain will prove that the author is at the genesis of their work.

[0047] For a better understanding of the invention, a drawing is attached, of which [Fig.1] schematically represents said process according to the invention.

[0048] With reference to [Fig. 1]:

[0049] A method for proving copyright in an original primary work and, where secondary or derivative works generated by artificial intelligence are derived from that same work (1), is characterized in that the original creator creates a work or document or book or collection incorporating all information about the natural and / or legal person and an exact description of their creation or invention in accordance with the definition of documentary evidence or written evidence resulting from a sequence of letters, characters, numbers or any other signs or symbols having an intelligible meaning, regardless of their medium and transmission methods (2), then the complete document is time-stamped and recorded on the blockchain to produce a digital fingerprint (3) 128-bit, 160-bit, 256-bit or 512-bit hash of the input data, depending on the algorithm selected,and is intended for cryptographic purposes through a trusted third-party partner, the registration ensures the printing of an attestation or certificate (3a) and a reproducible QR code (3b) on the product(s) (3c) and the complete document (3d) for the purpose of verifying whether, during a scan (3e), it corresponds to the attestation and the digital hash.

[0050] According to a supplementary preference for drafting, a work or document or book or collection integrating all information about the natural and / or legal person and an exact description of their creation or invention in accordance with the definition of literal evidence, or written evidence resulting from a sequence of letters, characters, numbers or any other signs or symbols having an intelligible meaning, regardless of their medium and methods of transmission, a software application which is a tool for detecting text and / or a drawing and / or an illustration generated by a LA which will verify the complete document before its publication and / or registration that it is created by a human to avoid potential problems of plagiarism or manipulation of content by a LA and will produce a verification report may integrate the final document.

[0051] In short, the text is humanized and undetectable by 1'1.A and, depending on the circumstances, is combined with anti-copying or anti-plagiarism software, a solution that allows testing and examining the authenticity of textual content. Specifically, this type of tool can analyze a text by comparing it with all the files (publications, documents, articles, books available on the web) in its database. Thus, the anti-plagiarism software detects word, phrase, and lexical frequency matches and highlights duplicate content.

[0052] According to one embodiment of the invention, the author has the option of generating a QR code block at the end of the complete document drafting process and after the application of humanization and AI protection software has been run, and a QR code after its registration (time stamp and blockchain). It will be possible to re-register the QR code block on the blockchain.

[0053] It is perfectly understandable and acceptable that the invention may integrate a private and / or public secure blockchain device.

[0054] If the author wishes to register his invention with a public body, or a regulated profession, the latter shall affix to all these paper or electronic documents the author's final QR code incorporating the various parameters of the process of the invention.

[0055] It will be up to the author to affix to the desired medium the QR code which will attest to his authorship even if secondary works are generated by a LA.

[0056] The object of the invention is to define and prove that the author is properly identified.

[0057] Whether or not the author uses AI for his creations is no longer relevant since the process satisfies upstream what is at the origin of the work or creation, whatever it may be.

[0058] Indeed, the method according to the invention initially proves the existence of an author, whether it be an author without the use of an LA or an author using an LA, who must also and previously identify themselves and describe precisely what their LA consists of. creation. Provide a precise description of your creation, including how you came up with the idea, a detailed description of the creation, and its purpose.

[0059] Moreover, 1'1. A cannot produce a creation or work solely on its own without the intervention of a natural person, at a minimum to launch the algorithmic calculation of the AI

[0060] The QR code is affixed to the product in an immutable and unalterable manner, using a hash value to reliably identify the author's file and provide sufficient information to locate the file's source, download it, and verify its content. Thus, the QR code derived from a hash value allows for the rapid retrieval of an element from a hash table, and cryptographic hash functions are particularly useful in contexts where it is necessary to protect against the risk of falsification (the creation of data with the same hash value as the intended data) by potentially malicious actors.

[0061] According to the present invention, in order to strengthen and improve security and combat hacking and to prevent AI from modifying or learning to generate more elaborate copies from an original creation, the QR-Code materialized on the registration document is available in several forms.

[0062] The standard QR code and / or the lenticular QR code which will allow the display of a second QR code and / or the QR code integrates in the central base the digital hash signature and / or a double QR code of which at least is invisible and revealed by ultraviolet reading.

[0063] According to a variant of the invention, the method (1) includes a software application capable of deforming the images and texts of the document (2) before they are loaded onto the web in order to make them counterproductive for training image-generating AIs.

[0064] This written description uses examples to disclose embodiments of the invention and also to enable any person skilled in the art to carry out embodiments of the invention, including the manufacture and use of any device or system and the performance of any incorporated process. The patentable scope of the invention is defined by the claims and may include other examples that are apparent to a person skilled in the art. These other examples are intended to fall within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements with non-substantial differences from the literal language of the claims.

[0065] Of course, the present invention is not limited to the embodiments or fields of activity previously described.

Claims

Demands

1. A method for proving copyright in an original primary work and, where secondary or derivative works generated by artificial intelligence are derived from that same work (1), characterized in that the original creator creates a work or document or book or collection incorporating all the information of a natural and / or legal person and an exact description of their creation or invention in accordance with the definition of documentary evidence or written evidence resulting from a sequence of letters, characters, numbers or any other signs or symbols having an intelligible meaning, regardless of their medium and method of transmission (2), then the document constituting the complete work is time-stamped and recorded on the blockchain to produce a digital fingerprint (3) 128-bit, 160-bit, 256-bit or 512-bit hash of the input data, depending on the algorithm selected,and is intended for cryptographic purposes through a trusted third-party partner, the registration ensures the printing of an attestation or certificate (3a) and a reproducible QR code (3b) on product(s) (3c) and the complete document (3d) for the purpose of verifying whether, during a scan (3e), it corresponds to the attestation and the digital hash.

2. A method for proving copyright in an original primary work and when secondary or derivative works generated by artificial intelligence are derived from this same work according to claim 1, characterized in that a software application is a tool for detecting text and / or drawing and / or illustration generated by an AI which will verify the document constituting the work before its publication and / or registration, whether it is created by a human, to avoid potential problems of plagiarism or manipulation of content by an AI and will produce a verification report which integrates the final document constituting the work.

3. A method for proving copyright in an original primary work and, where secondary or derivative works generated by artificial intelligence are derived from that same work according to claims 1-2, characterized in that the human text is undetectable by AI through the use of anti-copy or anti-plagiarism software to test and examine the authenticity of textual content and that this type of tool has the ability to analyze a text, comparing it with all the files (publications, documents, articles, books present on the web) in its database and detects word, phrase and lexical frequency matches, and highlights duplicate content.

4. A method for proving copyright in an original primary work and when secondary or derivative works generated by artificial intelligence are derived from this same work according to claims 1-2-3, characterized in that a QR-Code block is used at the end of the drafting of the complete document constituting the work and after the AI ​​protection software application has run, producing a QR-code after its recording (time stamp and blockchain) and then the QR-code block is re-recorded on the Blockchain.

5. A method for proving copyright in an original primary work and when secondary or derivative works generated by artificial intelligence are derived from this same work according to one of the preceding claims, characterized in that a final QR code is affixed to paper or electronic documents of the author according to the method of the invention.

6. A method for proving copyright in an original primary work and where secondary or derivative works generated by artificial intelligence are derived from that same work according to one of the preceding claims, characterized in that the author of the primary work must first identify himself and describe precisely what his creation consists of and then provide a precise description of his creation which includes how he came up with the idea, a precise description of the creation and for what purpose.

7. A method for proving copyright in an original primary work and, where secondary or derivative works generated by artificial intelligence are derived from that same work according to any one of the preceding claims, characterized in that the QR code is affixed to a product in an immutable and unalterable manner, using a hash value to reliably identify the file corresponding to the author's primary work by providing sufficient information to locate the sources of the file, download it and verify its contents and that said QR code from a hash value allows the quick consultation of an element in a hash table and that cryptographic hash functions are suitable in contexts where it is necessary to protect against the risks of falsification (the creation of data with the same hash value as the expected data) by potentially malicious participants.

8. A method for proving copyright in an original primary work and when secondary or derivative works generated by artificial intelligence are derived from this same work according to one of the preceding claims, characterized in that the QR-Code materialized on the primary work applied to the recorded document is available in several forms, namely a standard QR-Code and / or a lenticular QR-Code which will display a second QR-Code and / or a QR-Code which integrates the digital hash signature in the central base and / or a double QR-Code of which at least one is invisible and revealed by ultraviolet reading.

9. A method for proving copyright in an original primary work and where secondary or derivative works generated by artificial intelligence are derived from that same work according to one of the preceding claims, characterized in that the method (1) includes a software application capable of distorting the images and texts of the document (2) before they are uploaded to the web in order to make them counterproductive for training image-generating AIs.

Images