Adding watermarks to artificial intelligence (AI) generated compositions based on identification of sources

Abstract

An Artificial Intelligence (AI) generated composition is retrieved. For example, an AI generated composition may be an image generated by an AI algorithm. The AI generated composition is generated based on a training corpus that was used to train the AI algorithm. The training corpus comprises a plurality compositions (e.g., ten thousand images). A similarity algorithm is used to identify a snippet of the AI generated composition that matches a snippet of the one of the plurality of compositions. A watermark associated with the one of the plurality of compositions is identified. The watermark is inserted into the AI generated composition. The watermark may be then used to track illegal copies of the AI generated composition.

Description

Atty. Docket: 92053151 Adding Watermarks to Artificial Intelligence (AI) Generated Compositions Based on Identification of Sources FIELD

[0001] The disclosure relates generally to using watermarks to track compositions and particularly to using watermarks to track AI generated compositions. BACKGROUND

[0002] With Artificial Intelligence (AI) generated compositions (e.g., an AI generated image), existing watermarks in the training corpus are likely no longer detectable due to how Large Language Models (LLMs) AI algorithms work. In addition, because of how LLMs work, the ability to identify the sources used to generate the AI generated composition cannot be accomplished using the LLM. This is because Large Language Models (LLMs) may have thousands of layers. The information between the layers is not passed to the lower-level layers, thus there is no way to accurately identify the sources used to generate the AI generated composition based on the LLM. Because of these two issues, it is currently not possible to accurately watermark AI generated compositions. SUMMARY

[0003] These and other needs are addressed by the various embodiments and configurations of the present disclosure. The present disclosure can provide a number of advantages depending on the particular configuration. These and other advantages will be apparent from the disclosure contained herein.

[0004] An Artificial Intelligence (AI) generated composition is retrieved. For example, an AI generated composition may be an image generated by an AI algorithm. The AI generated composition is generated based on a training corpus that was used to train the AI algorithm. The training corpus comprises a plurality compositions (e.g., ten thousand images). A similarity algorithm is used to identify a snippet of the AI generated composition that matches a snippet of the one of the plurality of compositions. A watermark associated with the one of the plurality of compositions is identified. The watermark is inserted into the AI generated composition. The watermark may be then used to track illegal copies of the AI generated composition.Atty. Docket: 92053151

[0005] The phrases "at least one", "one or more", “or,” and "and / or" are open-ended expressions that are both conjunctive and disjunctive in operation. For example, each of the expressions "at least one of A, B and C", "at least one of A, B, or C", "one or more of A, B, and C", "one or more of A, B, or C", "A, B, and / or C", and "A, B, or C" means A alone, B alone, C alone, A and B together, A and C together, B and C together, or A, B and C together.

[0006] The term "a" or "an" entity refers to one or more of that entity. As such, the terms "a" (or "an"), "one or more" and "at least one" can be used interchangeably herein. It is also to be noted that the terms “comprising,” “including,” and “having” can be used interchangeably.

[0007] The term “automatic” and variations thereof, as used herein, refers to any process or operation, which is typically continuous or semi-continuous, done without material human input when the process or operation is performed. However, a process or operation can be automatic, even though performance of the process or operation uses material or immaterial human input, if the input is received before performance of the process or operation. Human input is deemed to be material if such input influences how the process or operation will be performed. Human input that consents to the performance of the process or operation is not deemed to be “material.”

[0008] Aspects of the present disclosure may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, micro-code, etc.) or an embodiment combining software and hardware aspects that may all generally be referred to herein as a “circuit,” “module” or “system.” Any combination of one or more computer readable medium(s) may be utilized. The computer readable medium may be a computer readable signal medium or a computer readable storage medium.

[0009] A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or FlashAtty. Docket: 92053151 memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device.

[0010] A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0011] The terms “determine,” “calculate” and “compute,” and variations thereof, as used herein, are used interchangeably, and include any type of methodology, process, mathematical operation, or technique.

[0012] The term “means” as used herein shall be given its broadest possible

[0014] The preceding is a simplified summary to provide an understanding of some aspects of the disclosure. This summary is neither an extensive nor exhaustive overview of the disclosure and its various embodiments. It is intended neither toAtty. Docket: 92053151 identify key or critical elements of the disclosure nor to delineate the scope of the disclosure but to present selected concepts of the disclosure in a simplified form as an introduction to the more detailed description presented below. As will be appreciated, other embodiments of the disclosure are possible utilizing, alone or in combination, one or more of the features set forth above or described in detail below. Also, while the disclosure is presented in terms of exemplary embodiments, it should be appreciated that individual aspects of the disclosure can be separately claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Fig. 1 is a block diagram of a first illustrative system for adding watermarks to AI generated compositions.

[0016] Fig. 2 is a block diagram of a second illustrative system for adding watermarks to AI generated compositions.

[0017] Fig. 3 is a flow diagram of a process for adding watermarks to AI generated compositions.

[0018] Fig. 4 is a flow diagram of a process for identifying watermarks in a training corpus.

[0019] Fig. 5 is a flow diagram of a process for identifying watermarks in AI generated compositions.

[0020] Fig. 6A is a diagram of a watermark that identifies an original copyright owner and a percentage of contribution of the original copyright owner.

[0021] Fig. 6B is a diagram of a plurality of watermarks that identifies a plurality original copyright owners and a plurality percentages of contribution of the plurality of original copyright owners.

[0022] Fig. 6C is a diagram of a watermark that identifies an original copyright owner, a percentage of contribution by the original copyright owner, and current owner.

[0023] Fig. 6D is a diagram of a plurality of watermarks that identify a plurality of original copyright owners, a plurality of percentages of contribution by the original copyright owners and a current copyright owner.

[0024] Fig. 6E is a diagram of a watermark that identifies an original copyright owner, a percentage of contribution by the original copyright owner, and a chain-of- title.Atty. Docket: 92053151

[0025] Fig. 6F is a diagram of plurality of watermarks that identify a plurality of original copyright owners, a plurality of percentages of contribution by the original copyright owners, and a chain-of-title.

[0026] Fig. 6G is a diagram of watermark that comprises an AI algorithm identifier, one or more original copyright owner(s), one or more percentages of contribution by the original copyright owner(s), and a current owner or a chain-of-title.

[0027] Fig. 7 is a flow diagram of a process for identifying watermarks and / or portions of watermarks in an AI generated composition.

[0028] In the appended figures, similar components and / or features may have the same reference label. Further, various components of the same type may be distinguished by following the reference label by a letter that distinguishes among the similar components. If only the first reference label is used in the specification, the description is applicable to any one of the similar components having the same first reference label irrespective of the second reference label. DETAILED DESCRIPTION

[0029] Fig. 1 is a block diagram of a first illustrative system 100 for adding watermarks to AI generated composition(s) 128. The first illustrative system 100 comprises communication devices 101A-101N, a network 110, and a watermarking server 120.

[0030] The communication devices 101A-101N can be or may include any user communication device 101 that can communicate on the network 110, such as a Personal Computer (PC), a cellular telephone, a Personal Digital Assistant (PDA), a tablet device, a notebook device, a laptop computer, a smartphone, and / or the like. As shown in Fig. 1, any number of communication devices 101A-101N may be connected to the network 110, including only a single communication device 101.

[0031] The network 110 can be or may include any collection of communication equipment that can send and receive electronic communications, such as the Internet, a Wide Area Network (WAN), a Local Area Network (LAN), a packet switched network, a circuit switched network, a cellular network, a combination of these, and the like. The network 110 can use a variety of electronic protocols, such as Ethernet, Internet Protocol (IP), Hyper Text Transfer Protocol (HTTP), Web Real-Time Protocol (Web RTC), and / or the like. Thus, the network 110 is an electronicAtty. Docket: 92053151 communication network configured to carry messages via packets and / or circuit switched communications.

[0032] The watermarking server 120 can be or may include any hardware coupled with software that can be used to manage the use of watermarks. In one embodiment, the watermarking server 120 may be a communication device 101. The watermarking server 120 comprises music 121, videos / images 122, documents 123, source code 124, a watermark identifier / filter 125, a training corpus 126, an AI algorithm 127, AI generated compositions(s) 128, a watermark inserter 129, watermarked AI generated composition(s) 130, a similarity algorithm 131, and a watermark database 132.

[0033] The music 121 can be any type of music / audio information that is stored in electronic form, such as music 121 stored on a disk, music 121 stored in a MP3 file, music in a WAV file, music 121 stored using an Advance Audio Coding (AAC), and / or the like. The music 121 may comprise other types of audio information, such as a podcast, audio books, and / or the like. The music 121 may be owned by one or more copyright owners.

[0034] The videos / images 122 may be any type of video / image that is stored in electronic form, such as MP4, MPEG, MOV, AVI, and / or the like. The videos / images 122 may be owned by one or more copyright owners.

[0035] The documents 123 may be any type of document 123 in any format, such as a Word® document 123, a PDF document 123, a text document 123, a spreadsheet, a presentation document 123, and / or the like. The documents 123 may be owned by one or more copyright owners.

[0036] The source code 124 may be any type of software / firmware, such as open- source software, proprietary software, third-party software, and / or the like. The source code 124 may be written in various types of programming languages, such as C, C++, Java, Python, Cobol, JavaScript, Hypertext Text Markup Language (HTML), and / or the like. The source code 124 may be binary source code 124, machine language source code 124, and / or the like.

[0037] The watermark identifier / filter 125 may be any hardware coupled with software that can identify watermarks a composition. The watermark identifier / filter 125 may identify watermarks in the training corpus 126, in the watermarked AI generated composition(s) 130, and / or the like. The watermark identifier / filter 125 may identify watermarks using various watermarking techniques. The watermarkAtty. Docket: 92053151 identifier / filter 125 may identify watermarks based on the watermarks in the watermark database 132. The watermark identifier / filter 125 may be used by a user to filter out specific types of music 121, videos / images 122, documents 123, source code 124, and / or the like to create the training corpus 126.

[0038] The training corpus 126 may be any type of training corpus 126, such as a music / audio training corpus 126, a video training corpus 126, an image training corpus 126, a document training corpus 126, a source code training corpus 126, and / or the like. The training corpus 126 may or may not have watermarks. The training corpus 126 may comprise a plurality difference compositions. For example, the training corpus 126 may comprise a thousand images.

[0039] The AI algorithm 127 may be any type of AI algorithm 127 that can generate AI generated composition(s) 128, such as ChatGPT, Chatsonic, Botsonice, OpenAI playground, Claude, Bard AI, and / or the like. The AI algorithm 127 is trained on the training corpus 126. The AI algorithm 127 generates the AI generated composition(s) 128 based on the training corpus 126.

[0040] The AI generated composition(s) 128 are compositions that are generated as a result of input to the AI algorithm 127. The input may be user input, automated input (e.g., from another AI algorithm, and / or the like). For example, the input may be to create an image based on the artist Monet where a person is sitting overlooking a lake. The AI generated composition(s) 128 may be based on one or more of the music 121, the videos / images 122, the documents 123, the source code 124, and / or the like.

[0041] The watermark inserter 129 can be or may include any hardware coupled with software that can insert watermarks into the AI generated composition(s) 128. The watermark inserter 129 may insert the watermark(s) using various processes, such as in specific location(s), based on the AI generated composition 128 (e.g., in different paces based on the AI composition type), based on a media type, and / or the like. The watermark inserter 129 may place the watermark(s) into the AI generated composition 128 based on similar pixels / sounds, in metadata (e.g., for a Word® document 123) to obfuscate the watermark(s). For source code 124, the watermark inserter 129 may add spaces and / or comments in specific rows of a source code file. For binaries, the watermark inserter may insert the watermark in padding, at the beginning or end of the binary, in metadata associated with the binary, and / or the like.Atty. Docket: 92053151

[0042] The watermarked AI generated composition(s) 130 are AI generated composition(s) 128 that have an added watermark(s) inserted by the watermark inserter 129.

[0043] The similarity algorithm 131 can be any type of algorithm that can be used to compare compositions in the training corpus 126 to the AI generated composition(s) 128. For example, a vector AI algorithm can take snippets of compositions in the training corpus 126 and then converts the snippets into vectors (e.g., floating point vectors). The vectors are compared that to vectors generated from snippets of the AI generated composition 128 to determine exact matches (e.g., a same vector) or similar matches (e.g., a close vector). Another alternative is to compare hashes of snippets of the AI generated composition 128 to hashes of snippets of compositions of the training corpus 126. Another option would be to directly compare the snippets of the compositions of the training corpus 126 to snippets of the AI generated composition 128.

[0044] The watermark database 132 is a database of watermarks that are associated with compositions of the training corpus 126. For example, if the training corpus 126 comprises ten images from ten different artists, there may be a watermark associated with each of the ten different artists stored in the watermark database 132. The compositions in the training corpus may each have a corresponding watermark.

[0045] Fig. 2 is a block diagram of a second illustrative system 200 for adding watermarks to AI generated compositions 128. The music 121, the videos / images 122, the documents 123, and / or the source code 124 are inputs 121-124 to the watermark identifier / filter 125. A user (via user input 201) identifies the specific inputs 121-124 to select / filter using the watermark identifier / filter 125. Once the specific inputs 121-124 (e.g., a group of files that comprise the training corpus 126) are selected, the watermark identifier / filter 125 identifies any watermarks in the training corpus 126. The watermark identifier / filter 125 can identify the watermarks using the watermark database 132, using known patterns / positions, and / or the like.

[0046] The training corpus 126 is then used to train the AI algorithm 127. Based on user input 201 (or an automated input), the AI algorithm 127 generates an AI generated composition 128. The AI generated composition 128 is input into the similarity algorithm 131 along with the training corpus 126 to identify matches / similarities. For example, the similarity algorithm 131 may be a vector AIAtty. Docket: 92053151 algorithm that breaks the training corpus 126 / AI generated composition 128 into snippets and then generates vectors (e.g., floating point vectors) for the snippets that are compared to each other. For example, if there were 10,000 images 122 in the training corpus 126 that were used to train the AI algorithm 127, the similarity algorithm 131 can identify the likely percentages of which of the 10,000 images 122 in the AI generated composition 128 (in this example an image) came from based on comparing the vectors.

[0047] Based on the percentages, the watermarks are added to the AI generated composition 128. For example, there may be 2% associated with image A, 1.2% associated with image B, and .62% associated with image C. The watermark inserter 129 may use a threshold (e.g., that may be defined via the user input 201). For example, the threshold may be .5%, so any images that have a percentage below .5% likelihood score are not identified. The watermark inserter 129 takes the percentages and identifies the watermark(s) in the watermark database 132 that are associated with the specific files in the training corpus 126 that match or are similar to the snippets in the AI generated composition 128. Based on the identified files (e.g., images), the watermark inserter 129 inserts the respective watermark(s) (e.g., as described in Figs. 6A-6G) into the AI generated composition 128 to produce the watermarked AI generated composition 130.

[0048] The watermarks may be placed into the AI generated composition(s) 128 using a known algorithm that places the watermark(s) in specific areas in the AI generated composition 128. For example, the watermarks may be placed in specific locations in an image, at specific times in music 121, in specific frames of a video, using comments in source code 124, and / or the like. If an identified watermarked file that was identified in the training corpus 126 (which may be any type) is not identified by the similarity algorithm 131 as having a corresponding snippet of the AI generated composition 128, the watermark is inserted by the watermark inserter 129.

[0049] Fig. 3 is a flow diagram of a process for adding watermarks to AI generated composition(s) 128. Illustratively, the communication devices 101A-101N, the watermarking server 120, the music 121, the videos / images 122, the documents 123, the source code 124, the watermark identifier / filter 125, the training corpus 126, the AI algorithm 127, the AI generated composition(s) 128, the watermark inserter 129, the watermarked AI generated composition(s) 130, the similarity algorithm 131, andAtty. Docket: 92053151 the watermark database 132 are stored-program-controlled entities, such as a computer or microprocessor, which performs the method of Figs. 3-7 and the processes described herein by executing program instructions stored in a computer readable storage medium, such as a memory (i.e., a computer memory, a hard disk, and / or the like). Although the methods described in Figs. 3-7 are shown in a specific order, one of skill in the art would recognize that the steps in Figs. 3-7 may be implemented in different orders and / or be implemented in a multi-threaded environment. Moreover, various steps may be omitted or added based on implementation.

[0050] The process starts in step 300. The similarity algorithm 131 retrieves the AI generated composition 128 in step 302. The similarity algorithm 131 identifies snippet(s) of the AI generated composition 128 that match or are similar to snippet(s) of the compositions of the training corpus 126 in step 304. For example, the similarity algorithm 131 may use vectors of snippets, hashes of snippets, and / or snippets to compare between the AI generated composition 128 and the compositions in the training corpus 126.

[0051] If there are no matches / similarities in step 306, the process goes to step 314. Otherwise, if there are one or more matches / similarities in step 306, the watermark inserter 129 identifies the percentages of the identified snippets compared to the compositions in the training corpus 126 in step 308. For example, if the total disk space of the compositions in the training corpus 126 was ten megabytes and the size of the identified snippet was ten kilobytes, the percentage would be .1%. If there are multiple snippets identified from the same composition or from other compositions owned by the same copyright owner, the percentages are added. For example, if the percentage for snippet A from composition A (owned by copyright owner A) was .1% and the percentage for snippet B from composition B (also owned by copyright owner A) was also .1%, the percentage would be .2% for copyright owner A.

[0052] The watermark inserter 129 identifies, in step 310, watermark(s) associated with the matched / similar snippets of the compositions of the training corpus 126. The watermark inserter 129 inserts, in step 312, the watermark(s) (e.g., those described in Figs. 6A-6G) into the AI generated composition 128.Atty. Docket: 92053151

[0053] The process determines, in step 314, if the process is complete. If the process is not complete in step 314, the process goes back to step 302. Otherwise, if the process is complete in step 314, the process ends in step 316.

[0054] Fig. 4 is a flow diagram of a process for identifying watermarks in a training corpus 126. The process stars in step 400. The AI algorithm 127 determines, in step 402, if the AI algorithm 127 is to be trained. If the AI algorithm 127 is not to be trained in step 402, the process of step 402 repeats.

[0055] Otherwise, if the AI algorithm 127 is to be trained in step 402, the AI algorithm 127 gets the training corpus 126 in step 404. The watermark identifier 125 identifies any watermark(s) in the training corpus 126 in step 406. The watermark identifier 125 may identify the watermark(s) in the training corpus 126 in various ways, such as using known locations, using an identification algorithm, and / or the like. The identified watermark(s) are stored in the watermark database 132 in step 408. Although not shown in Fig. 3, the watermark identifier 125 may remove the watermark between steps 406 and 408.

[0056] The process determines, in step 410, if the process is complete. If the process is not complete in step 410, the process goes back to step 402. Otherwise, if the process is complete in step 410, the process ends in step 412.

[0057] Fig. 5 is a flow diagram of a process for identifying watermarks in AI generated compositions 128. The process starts in step 500. The watermark identifier 125 determines, in step 502, if there is a watermarked AI composition 130 to review. If there is not a watermarked composition to review in step 502, the process of step 502 repeats. Otherwise, the watermark identifier 125 determines if the watermarked AI composition 130 contains known watermarks in step 504. The known watermarks may be from the watermark database 132.

[0058] If there are not any known watermarks in step 506, the watermarked AI composition is identified as not containing any watermarks in step 506 and the process goes to step 510. If the watermarked AI composition 130 contains one or more watermarks in step 504, the watermark(s) are identified and saved in step 508. The watermark(s) and any associated information are displayed in step 510. For example, the information in the watermarks 600A-600G as described in Figs. 6A-6G may be displayed to a user in step 510. Although not shown, step 510 may includeAtty. Docket: 92053151 validating if the watermark is valid or an illegal copy of the AI generated composition 128.

[0059] The process determines, in step 512, if the process is complete. If the process is not complete in step 512, the process goes back to step 502. Otherwise, if the process is complete in step 512, the process ends in step 514.

[0060] Fig. 6A is a diagram of a watermark 600A that identifies an original copyright owner(s) and a percentage of contribution of the original copyright owner(s). The watermark 600A is associated with one or more compositions of the training corpus 126. There may be more than one copyright owner if the composition of the training corpus 126 has multiple associated copyright owners. For example, if a single musical composition X in the training corpus 126 has two copyright owners and is identified based on a snippet in the AI generated composition 128, the watermark 600A will also have the two owners with the same percentage. The original copyright owner(s) is the copyright owner(s) of the copyright associated with a composition(s) in the training corpus 126 when the AI algorithm 127 is trained. This applies to Figs. 6A-6G. For example, if a snippet of image A in the training corpus 126 is matched / similar to a snippet in the AI generated composition 128, the original copyright owner in the watermark 600A would be the copyright owner of the image A.

[0061] The percentage is the amount that the identified snippet is in relation to the total training corpus 126 like described above. This also applies to Figs. 6A-6G.

[0062] Fig. 6B is a diagram of a plurality of watermarks 600B that identifies a plurality copyright owners and a plurality percentages of contribution of the plurality of copyright owners. For example, if there were two snippets identified where snippet A has a 1% match to composition A owned by copyright owner A and snippet N has a 2% similarity to composition N owned by copyright owner N the corresponding percentages would be 1% to copyright owner A and 2% to copyright owner N.

[0063] If there are multiple original owners that are identified by the similarity algorithm 131, the watermark may comprise multiple watermarks. The multiple watermarks may be placed in predefined locations. For example, a first watermark for the copyright owner A / percentage may be inserted at a first location and a second watermark for the copyright owner N / percentage may be placed in a second location. This could apply to any situation where there are multiple watermarks.Atty. Docket: 92053151

[0064] Fig. 6C is a diagram of a watermark 600C that identifies an original copyright owner, a percentage of contribution by the original copyright owner, and current owner. The current owner would be the current owner of the AI generated material 128.

[0065] Fig. 6D is a diagram of a plurality of watermarks 600D that identify a plurality of original copyright owner(s), a plurality of percentages of contribution by the original copyright owner(s) and a current copyright owner. The watermarks 600D are similar to the watermark 600C, but with multiple copyright owners / percentages.

[0066] Fig. 6E is a diagram of a watermark 600E that identifies an original copyright owner, a percentage of contribution by the original copyright owner, and a chain-of-title. The watermark 600E is similar to the watermark 600C, but instead of a current copyright owner, there is a chain-of-title. The chain-of-title shows all owners of the AI generated composition 128 (or it could be for a subset of previous owners of the AI generated composition 128). For example, the chain-of-title may comprise watermarks for each person who have owned the AI generated composition 128. The chain-of-title may include other information, such as a number of allowable sales, a maximum number of copies, and / or the like as described in US Patent Application No. 18 / 228228 titled “Using Watermarks to Identify a Chain of Title in Media,” which is incorporated herein by reference. The chain-of-title may be verified using a watermark tracking service.

[0067] Fig. 6F is a diagram of plurality of watermarks 600F that identify a plurality of original copyright owners, a plurality of percentages of contribution by the original copyright owners, and a chain-of-title. The difference between Fig. 6E and Fig. 6F is that the watermarks 600F has multiple original copyright owners.

[0068] Fig. 6G is a diagram of watermark 600G that comprises an AI algorithm identifier, one or more original copyright owner(s), one or more percentages of contribution by the original copyright owner(s), and a current owner or a chain-of- title. The difference between the watermark 600G and the watermarks 600A-600F is the addition of the AI algorithm identifier. In other words, the AI algorithm identifier may be added to any of the watermarks 600A-600F.

[0069] As one can envision, various combinations of the watermark(s) 600A- 600G / information in the watermarks 600A-600B may be used. For example, only theAtty. Docket: 92053151 current owner / chain-of-title may be in the watermark 600 or the watermark 600 may comprise only the AI algorithm identifier.

[0070] The watermark(s) 600 may be added based on pixels (i.e., using an indexing scheme or watermark pointers). The watermarks 600 may be used to prevent a vishing attack that is generated by another AI algorithm. For example, if a hacker takes multiple videos of a person and generates a fake video, the fake video will not have the watermark(s) 600 or have multiple watermarks 600 that are from separate videos. Another option would be to add a hash of the AI material that is tied to the watermark to prevent tampering.

[0071] In one embodiment, the watermark 600 may be added by the artist as part of the artist’s work with the AI algorithm 127. Another option would be to use this method for making payments to the copyright owners.

[0072] In one embodiment, this may be provided as part of a service where a blockchain is used to track the use of the watermarks 600.

[0073] Fig. 7 is a flow diagram of a process for identifying watermarks 600 and / or portions of watermarks 600 in an AI generated composition 128. The process starts in step 700. The similarity algorithm 131 retrieves the AI generated composition 128 in step 702. The similarity algorithm 131 retrieves the watermark(s) 600 identified in the training corpus 126 in step 704 (e.g., the watermark(s) 600 identified in step 406).

[0074] The similarity algorithm 131 searches the AI generated composition 128 for the identified watermark(s) 600 and / or portions of the identified watermark(s) 600 in step 706. The similarity algorithm 131 may identify the watermark(s) 600 based on various factors, such as, location (e.g., a location in an image, a line in source code, etc.), how much of a match of the watermark 600, data around the watermark(s) 600, and / or the like.

[0075] If there are not any match(es) / identified portion(s) in step 708, the process goes to step 716. Otherwise, the similarity algorithm 131 determines, in step 710, if the identified watermark(s) 600 / portion(s) of the watermark(s) are identified in the matched snippets (e.g., those matched in step 306). Step 710 may identify some portion(s) / matches that are in the snippets and some that are not. If there is one or more watermark(s) 600 / portions of watermark(s) that are identified in the matched snippets in step 710, the process goes to step 716. If there is one or more portions of watermark(s) identified in step 710 that are not in the identified snippets, theAtty. Docket: 92053151 portion(s) of the watermark(s) are removed from the AI generated composition 128 in step 712. The full watermark(s) are also added in step 312. The percentages (e.g., those identified in step 308) are updated, in step 714, based on the snippet size in a similar manner as described in Fig. 3 and the process goes to step 716.

[0076] The process determines, in step 716, if the process is complete. If the process is not complete in step 716, the process goes to step 702. Otherwise, if the process is complete in step 716, the process ends in step 718.

[0077] Examples of the processors as described herein may include, but are not limited to, at least one of Qualcomm® Snapdragon® 800 and 801, Qualcomm® Snapdragon® 610 and 615 with 4G LTE Integration and 64-bit computing, Apple® A7 processor with 64-bit architecture, Apple® M7 motion coprocessors, Samsung® Exynos® series, the Intel® Core™ family of processors, the Intel® Xeon® family of processors, the Intel® Atom™ family of processors, the Intel Itanium® family of processors, Intel® Core® i5-4670K and i7-4770K 22nm Haswell, Intel® Core® i5- 3570K 22nm Ivy Bridge, the AMD® FX™ family of processors, AMD® FX-4300, FX-6300, and FX-835032nm Vishera, AMD® Kaveri processors, Texas Instruments® Jacinto C6000™ automotive infotainment processors, Texas Instruments® OMAP™ automotive-grade mobile processors, ARM® Cortex™-M processors, ARM® Cortex-A and ARM926EJ-S™ processors, other industry- equivalent processors, and may perform computational functions using any known or future-developed standard, instruction set, libraries, and / or architecture.

[0078] Any of the steps, functions, and operations discussed herein can be performed continuously and automatically.

[0079] However, to avoid unnecessarily obscuring the present disclosure, the preceding description omits a number of known structures and devices. This omission is not to be construed as a limitation of the scope of the claimed disclosure. Specific details are set forth to provide an understanding of the present disclosure. It should however be appreciated that the present disclosure may be practiced in a variety of ways beyond the specific detail set forth herein.

[0080] Furthermore, while the exemplary embodiments illustrated herein show the various components of the system collocated, certain components of the system can be located remotely, at distant portions of a distributed network, such as a LAN and / or the Internet, or within a dedicated system. Thus, it should be appreciated, that theAtty. Docket: 92053151 components of the system can be combined in to one or more devices or collocated on a particular node of a distributed network, such as an analog and / or digital telecommunications network, a packet-switch network, or a circuit-switched network. It will be appreciated from the preceding description, and for reasons of computational efficiency, that the components of the system can be arranged at any location within a distributed network of components without affecting the operation of the system. For example, the various components can be located in a switch such as a PBX and media server, gateway, in one or more communications devices, at one or more users’ premises, or some combination thereof. Similarly, one or more functional portions of the system could be distributed between a telecommunications device(s) and an associated computing device.

[0081] Furthermore, it should be appreciated that the various links connecting the elements can be wired or wireless links, or any combination thereof, or any other known or later developed element(s) that is capable of supplying and / or communicating data to and from the connected elements. These wired or wireless links can also be secure links and may be capable of communicating encrypted information. Transmission media used as links, for example, can be any suitable carrier for electrical signals, including coaxial cables, copper wire and fiber optics, and may take the form of acoustic or light waves, such as those generated during radio-wave and infra-red data communications.

[0082] Also, while the flowcharts have been discussed and illustrated in relation to a particular sequence of events, it should be appreciated that changes, additions, and omissions to this sequence can occur without materially affecting the operation of the disclosure.

[0083] A number of variations and modifications of the disclosure can be used. It would be possible to provide for some features of the disclosure without providing others.

[0084] In yet another embodiment, the systems and methods of this disclosure can be implemented in conjunction with a special purpose computer, a programmed microprocessor or microcontroller and peripheral integrated circuit element(s), an ASIC or other integrated circuit, a digital signal processor, a hard-wired electronic or logic circuit such as discrete element circuit, a programmable logic device or gate array such as PLD, PLA, FPGA, PAL, special purpose computer, any comparableAtty. Docket: 92053151 means, or the like. In general, any device(s) or means capable of implementing the methodology illustrated herein can be used to implement the various aspects of this disclosure. Exemplary hardware that can be used for the present disclosure includes computers, handheld devices, telephones (e.g., cellular, Internet enabled, digital, analog, hybrids, and others), and other hardware known in the art. Some of these devices include processors (e.g., a single or multiple microprocessors), memory, nonvolatile storage, input devices, and output devices. Furthermore, alternative software implementations including, but not limited to, distributed processing or component / object distributed processing, parallel processing, or virtual machine processing can also be constructed to implement the methods described herein.

[0085] In yet another embodiment, the disclosed methods may be readily implemented in conjunction with software using object or object-oriented software development environments that provide portable source code that can be used on a variety of computer or workstation platforms. Alternatively, the disclosed system may be implemented partially or fully in hardware using standard logic circuits or VLSI design. Whether software or hardware is used to implement the systems in accordance with this disclosure is dependent on the speed and / or efficiency requirements of the system, the particular function, and the particular software or hardware systems or microprocessor or microcomputer systems being utilized.

[0086] In yet another embodiment, the disclosed methods may be partially implemented in software that can be stored on a storage medium, executed on programmed general-purpose computer with the cooperation of a controller and memory, a special purpose computer, a microprocessor, or the like. In these instances, the systems and methods of this disclosure can be implemented as program embedded on personal computer such as an applet, JAVA® or CGI script, as a resource residing on a server or computer workstation, as a routine embedded in a dedicated measurement system, system component, or the like. The system can also be implemented by physically incorporating the system and / or method into a software and / or hardware system.

[0087] Although the present disclosure describes components and functions implemented in the embodiments with reference to particular standards and protocols, the disclosure is not limited to such standards and protocols. Other similar standards and protocols not mentioned herein are in existence and are considered to be includedAtty. Docket: 92053151 in the present disclosure. Moreover, the standards and protocols mentioned herein and other similar standards and protocols not mentioned herein are periodically superseded by faster or more effective equivalents having essentially the same functions. Such replacement standards and protocols having the same functions are considered equivalents included in the present disclosure.

[0088] The present disclosure, in various embodiments, configurations, and aspects, includes components, methods, processes, systems and / or apparatus substantially as depicted and described herein, including various embodiments, sub combinations, and subsets thereof. Those of skill in the art will understand how to make and use the systems and methods disclosed herein after understanding the present disclosure. The present disclosure, in various embodiments, configurations, and aspects, includes providing devices and processes in the absence of items not depicted and / or described herein or in various embodiments, configurations, or aspects hereof, including in the absence of such items as may have been used in previous devices or processes, e.g., for improving performance, achieving ease and\or reducing cost of implementation.

[0089] The foregoing discussion of the disclosure has been presented for purposes of illustration and description. The foregoing is not intended to limit the disclosure to the form or forms disclosed herein. In the foregoing Detailed Description for example, various features of the disclosure are grouped together in one or more embodiments, configurations, or aspects for the purpose of streamlining the disclosure. The features of the embodiments, configurations, or aspects of the disclosure may be combined in alternate embodiments, configurations, or aspects other than those discussed above. This method of disclosure is not to be interpreted as reflecting an intention that the claimed disclosure requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment, configuration, or aspect. Thus, the following claims are hereby incorporated into this Detailed Description, with each claim standing on its own as a separate preferred embodiment of the disclosure.

[0090] Moreover, though the description of the disclosure has included description of one or more embodiments, configurations, or aspects and certain variations and modifications, other variations, combinations, and modifications are within the scope of the disclosure, e.g., as may be within the skill and knowledge of those in the art,Atty. Docket: 92053151 after understanding the present disclosure. It is intended to obtain rights which include alternative embodiments, configurations, or aspects to the extent permitted, including alternate, interchangeable and / or equivalent structures, functions, ranges or steps to those claimed, whether or not such alternate, interchangeable and / or equivalent structures, functions, ranges or steps are disclosed herein, and without intending to publicly dedicate any patentable subject matter.

Claims

Atty. Docket: 92053151 CLAIMS What is claimed is:

1. A system comprising: a microprocessor; and a computer readable medium, coupled with the microprocessor and comprising microprocessor readable and executable instructions that, when executed by the microprocessor, cause the microprocessor to: retrieve an Artificial Intelligence (AI) generated composition, wherein the AI generated composition is generated based on a training corpus that was used to train an AI algorithm and wherein the training corpus comprises a plurality compositions; identify, using a similarity algorithm, a snippet of the AI generated composition that matches or is similar to a snippet of the one of the plurality of compositions; identify a watermark associated with the one of the plurality of compositions; and insert the watermark associated with the one of the plurality of compositions into the AI generated composition.

2. The system of claim 1, wherein identifying the watermark associated with the one of the plurality of compositions comprises identifying the watermark associated with the one of the plurality of compositions in the training corpus.

3. The system of claim 1, wherein identifying the watermark associated with the one of the plurality of compositions comprises identifying the watermark associated with the one of the plurality of compositions in a watermark database.

4. The system of claim 1, wherein the watermark associated with the one of the plurality of compositions inserted into the AI generated composition comprises an identifier of a copyright owner and a percentage of contribution of the copyright owner.

5. The system of claim 4, wherein the watermark associated with the one of the plurality of compositions inserted into the AI generated composition further comprises an identifier of a current owner.Atty. Docket: 92053151 6. The system of claim 4, wherein the watermark associated with the one of the plurality of compositions inserted into the AI generated composition further comprises a chain-of-title of the AI generated composition.

7. The system of claim 1, wherein the similarity algorithm identifies a plurality of snippets of the AI generated composition that match and / or are similar to a plurality of snippets of the plurality of compositions, wherein identifying the watermark associated with the one of the plurality of compositions comprises identifying a plurality of watermarks associated with the plurality of compositions, and wherein inserting the watermark associated with the one of the plurality of compositions into the AI generated composition comprises inserting the plurality of watermarks associated with the plurality of compositions into the AI generated composition.

8. The system of claim 7, wherein plurality of watermarks associated with the plurality of compositions are placed into the AI generated composition in separate locations within the AI generated composition.

9. The system of claim 7, wherein the plurality of watermarks associated with the plurality of compositions inserted into AI generated composition comprise a plurality identifiers of copyright owners and a plurality of percentages of contribution of the plurality of copyright owners.

10. The system of claim 9, wherein the plurality of watermarks associated with the one of the plurality of compositions inserted into the AI generated composition further comprises an identifier of a current owner.

11. The system of claim 9, wherein the plurality of watermarks inserted into the AI generated composition further comprises a chain-of-title of the AI generated composition.

12. The method of claim 1, wherein the similarity algorithm uses at least one of a vector, a hash, and the snippet of the one of the plurality of compositions.

13. The method of claim 1, wherein the watermark associated with the one of the plurality of compositions inserted into the AI generated compositionAtty. Docket: 92053151 comprises an identifier of an AI algorithm used to generate the AI generated composition.

14. The method of claim 1, wherein the microprocessor readable and executable instructions further cause the microprocessor to: identify, using the similarity algorithm, a portion of a second watermark in the AI generated composition, wherein the second watermark is in the training corpus; remove the portion of the second watermark from the AI generated composition; get the second watermark; and insert the second watermark into the AI generated composition.

15. A method comprising: retrieving, by a microprocessor, an Artificial Intelligence (AI) generated composition, wherein the AI generated composition is generated based on a training corpus that was used to train an AI algorithm and wherein the training corpus comprises a plurality compositions; identifying, by the microprocessor, using a similarity algorithm, a snippet of the AI generated composition that matches or is similar to a snippet of the one of the plurality of compositions; identifying, by the microprocessor, a watermark associated with the one of the plurality of compositions; and inserting, by the microprocessor, the watermark associated with the one of the plurality of compositions into the AI generated composition.

16. The method of claim 15, wherein identifying the watermark associated with the one of the plurality of compositions comprises identifying the watermark associated with the one of the plurality of compositions in the training corpus.

17. The method of claim 15, wherein the watermark associated with the one of the plurality of compositions inserted into the AI generated composition comprises an identifier of a copyright owner and a percentage of contribution of the copyright owner.Atty. Docket: 92053151 18. The method of claim 15, wherein the similarity algorithm identifies a plurality of snippets of the AI generated composition that match and / or are similar to a plurality of snippets of the plurality of compositions, wherein identifying the watermark associated with the one of the plurality of compositions comprises identifying a plurality of watermarks associated with the plurality of compositions, and wherein inserting the watermark associated with the one of the plurality of compositions into the AI generated composition comprises inserting the plurality of watermarks associated with the plurality of compositions into the AI generated composition.

19. The method of claim 18, wherein the plurality of watermarks associated with the plurality of compositions inserted into AI generated composition comprises a plurality identifiers of copyright owners and a plurality of percentages of contribution of the plurality of copyright owners.

20. A non-transient computer readable medium having stored thereon instructions that cause a processor to execute a method, the method comprising instructions to: retrieve an Artificial Intelligence (AI) generated composition, wherein the AI generated composition is generated based on a training corpus that was used to train an AI algorithm and wherein the training corpus comprises a plurality compositions; identify, using a similarity algorithm, a snippet of the AI generated composition that matches or is similar to a snippet of the one of the plurality of compositions; identify a watermark associated with the one of the plurality of compositions; and insert the watermark associated with the one of the plurality of compositions into the AI generated composition.

Images