WO2025150041A1

WO2025150041A1 - Privacy-preserving ai-based content generation

Info

Publication number: WO2025150041A1
Application number: PCT/IL2025/050020
Authority: WO
Inventors: Igor PECHERSKY
Original assignee: Anagog Ltd
Current assignee: Anagog Ltd
Priority date: 2024-01-08
Filing date: 2025-01-08
Publication date: 2025-07-17
Anticipated expiration: 2026-07-08

Abstract

The present disclosure provides a method of privacy-preserving content generation. The method includes extracting a set of characteristics of a user from data retained on a client device, encapsulating subsets of the extracted characteristics in a plurality of Artificial Intelligence (Al) ambassadors, using at least one of the Al ambassadors to receive content selected by a remote content generating computer based on the encapsulated characteristics, and presenting at least a portion of the received content to the user. The method may further include encapsulating at least one false characteristic of the user in at least one of the Al ambassadors and selecting the portion of the received content for presenting based on at least one extracted characteristic not encapsulated in the Al ambassador used for receiving the content.

Description

PRIVACY-PRESERVING AI-BASED CONTENT GENERATION

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U. S. Application No. 63/618,468, titled "Privacy-Preserving Artificial Intelligence Content Generation", filed Jan. 8, 2024, which is hereby incorporated by reference in its entirety without giving rise to disavowment.

FIELD OF INVENTION

[0002] The present disclosure relates to artificial intelligence-based content generation, and more particularly to privacy-preserving methods and systems for generating personalized content.

BACKGROUND

[0003] Artificial Intelligence (Al) systems have become increasingly sophisticated and prevalent in recent years, offering new possibilities for personalized content generation and user assistance in various creative and intellectual pursuits. These Al systems often rely on large amounts of user data to generate tailored content, recommendations, and experiences. However, this reliance on personal data has raised significant privacy concerns among users and regulators alike.

[0004] As the volume of content produced worldwide continues to escalate rapidly, individuals face growing challenges in effectively filtering and curating content that aligns with their interests. This has led to an increased demand for personalized Al assistance in content curation and creation. However, balancing the desire for personalized experiences with the need to protect user privacy has become a complex technical challenge.

[0005] Various approaches have been proposed to address privacy concerns in Al-driven personalization. These include techniques such as federated learning, differential privacy, and on-device processing. While these methods offer some level of privacy protection, they often come with trade-offs in terms of personalization accuracy, computational efficiency, or the granularity of user targeting.

[0006] Edge Al models have emerged as a promising direction for preserving user privacy by processing sensitive information directly on user devices. These models aim to understand multi-faceted patterns of user behavior without transmitting personal data to external servers. However, fully encapsulating an individual's personality and interests in a single Edge Al model on a device may raise its own set of privacy and security concerns.

[0007] Content generation networks, which involve collaborative groups of autonomous models or algorithms working together, have shown potential for creating personalized content. These networks may involve multiple models that share information about user preferences and generate relevant content based on that data. However, implementing such systems while maintaining strong privacy guarantees remains a significant challenge.

SUMMARY

[0008] According to an aspect of the present disclosure, a method of privacy-preserving content generation is provided. The method includes extracting a set of characteristics of a user from data retained on a client device. The method includes encapsulating subsets of the extracted characteristics in a plurality of Artificial Intelligence (Al) ambassadors. The method includes using at least one of the Al ambassadors to receive content selected by a remote content generating computer based on the encapsulated characteristics. The method includes presenting at least a portion of the received content to the user.

[0009] According to other aspects of the present disclosure, the method may include one or more of the following features. Encapsulating subsets of the extracted characteristics may comprise encapsulating at least one false characteristic of the user in at least one of the Al ambassadors. The method may further comprise selecting the portion of the received content for presenting to the user based on at least one extracted characteristic not encapsulated in the Al ambassador used for receiving the content. The data retained on the client device may comprise data extracted from readings of at least one sensor of the client device. The at least one sensor may comprise at least one of: a gyroscope, an accelerometer, a GPS receiver, a camera, a microphone, or a touchscreen sensor. The method may further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the plurality of Al ambassadors into a single format for presenting to the user. The single format may comprise a single webpage or a single document.

[0010] According to another aspect of the present disclosure, a system for privacypreserving content generation is provided. The system includes a client device comprising a processor and a memory storing instructions that, when executed by the processor, cause the client device to: extract a set of characteristics of a user from locally retained data; generate a plurality of Al ambassadors, each encapsulating a subset of the extracted characteristics; communicate with a remote content generating computer using at least one of the Al ambassadors to receive content selected based on the encapsulated characteristics; and present at least a portion of the received content to the user.

[0011] According to other aspects of the present disclosure, the system may include one or more of the following features. At least one of the Al ambassadors may encapsulate at least one false characteristic of the user. The instructions may further cause the client device to select the portion of the received content for presenting to the user based on at least one extracted characteristic not encapsulated in the Al ambassador used for receiving the content. The locally retained data may comprise data extracted from readings of at least one sensor of the client device. The instructions may further cause the client device to consolidate content received from the content generating computer using at least two Al ambassadors of the plurality of Al ambassadors into a single format for presenting to the user. The single format may comprise a single webpage or a single document.

[0012] According to another aspect of the present disclosure, a non-transitory computer- readable medium storing instructions is provided. When executed by a processor of a client device, the instructions cause the client device to perform a method of privacy-preserving content generation. The method includes modeling a personality representation of a user based on behavioral data stored on the client device. The method includes creating a plurality of Al ambassadors, each representing a partial aspect of the modeled personality representation. The method includes providing the Al ambassadors to a remote creative network for content generation, wherein the remote create network is configured to utilize the Al ambassadors or portion thereof to generate content. The method includes aggregating content received from the remote creative network to produce personalized content for the user.

[0013] According to other aspects of the present disclosure, the non-transitory computer- readable medium may include one or more of the following features. Creating the plurality of Al ambassadors may comprise encapsulating at least one false characteristic of the user in at least one of the Al ambassadors. The behavioral data may comprise data extracted from readings of at least one sensor of the client device. Aggregating the content received from the remote creative network may comprise applying a local key cipher to filter and recombine the content. Applying the local key cipher may comprise selecting content based on a determination of which Al ambassador best matches the modeled personality representation with respect to a specific aspect relating to a topic of the content.

[0014] According to another aspect of the present disclosure, a method of content generation is provided. The method comprises steps executed by a client device in communication with an Al based content generating computer. The steps include extracting a set of characteristics of a user of the client device from data retained on the client device. The steps include encapsulating a respective subset of the extracted set of characteristics of the user in each ambassador of a group of at least one Al ambassador. The steps include using at least one Al ambassador of the group, for receiving content selected by the content generating computer, the content being selected by the content generating computer based on the characteristics encapsulated in the Al ambassador. The steps include presenting at least a part of the received content selected by the content generating computer, to the user.

[0015] According to other aspects of the present disclosure, the method may include one or more of the following features. The group may comprise at least two Al ambassadors. The encapsulating may further comprise encapsulating at least one false characteristic of the user in at least one of the Al ambassadors of the group. The method may further comprise selecting the part of the received content for presenting. The method may further comprise selecting the part of the received content for presenting using at least one of the extracted characteristics of the user not encapsulated in the Al ambassador used for obtaining the content. The data retained on the client device may comprise data extracted from a reading of at least one sensor of the client device. The method may further comprise extracting the data retained on the client device from a reading of at least one sensor of the client device. The data stored on the client device may comprise data extracted from a reading of at least one sensor of the client device, the at least one sensor comprising at least one of the group consisting of: a gyroscope, a Wi-Fi receiver, a blue tooth receiver, a charger connection sensor, an accelerometer, a GPS receiver, a camera, a magnetometer, a proximity sensor, an ambient light sensor, a microphone, a touchscreen sensor, a fingerprint sensor, a pedometer, a barometer, a thermometer, and an air humidity sensor. The method may further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single format used in presenting. The method may further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single webpage used in presenting. The method may further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single document used in presenting.

[0016] According to another aspect of the present disclosure, a system for data distribution in a computer network is provided. The system comprises a circuit comprising a computer processor and a computer memory storing instructions that are executable by the computer processor, for performing steps. The steps include extracting a set of characteristics of a user of the client device from data retained on the client device. The steps include encapsulating a respective subset of the extracted set of characteristics of the user in each ambassador of a group of at least one Al ambassador. The steps include using at least one ambassador of the group, for receiving content selected by the content generating computer, the content being selected by the content generating computer based on the characteristics encapsulated in the Al ambassador. The steps include presenting at least a part of the received content selected by the content generating computer, to the user.

[0017] According to other aspects of the present disclosure, the system may include one or more of the following features. The group may comprise at least two Al ambassadors. The encapsulating may further comprise encapsulating at least one false characteristic of the user in at least one of the Al ambassadors of the group. The steps may further comprise selecting the part of the received content for presenting. The steps may further comprise selecting the part of the received content for presenting using at least one of the extracted characteristics of the user not encapsulated in the Al ambassador used for receiving the content. The data retained on the client device may comprise data extracted from a reading of at least one sensor of the client device. The steps may further comprise extracting the data retained on the client device from a reading of at least one sensor of the client device. The steps may further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single format used in presenting. The steps may further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single webpage used in presenting. The steps may further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single document used in presenting.

[0018] The foregoing general description of the illustrative embodiments and the following detailed description thereof are merely exemplary aspects of the teachings of this disclosure and are not restrictive.

BRIEF DESCRIPTION OF FIGURES

[0019] Non-limiting and non-exhaustive examples are described with reference to the following figures.

[0020] FIG. 1 illustrates a flowchart for a privacy-preserving content generation process, according to aspects of the present disclosure.

[0021] FIG. 2 depicts a flowchart of a method for privacy-preserving content generation, in accordance with example embodiments.

[0022] FIG. 3 shows a block diagram of a system for privacy-preserving content generation, according to an embodiment. [0023] FIG. 4 illustrates a block diagram of a privacy-preserving content generation system, according to aspects of the present disclosure.

[0024] FIG. 5 depicts a block diagram of another privacy-preserving content generation system, in accordance with example embodiments.

DETAILED DESCRIPTION

[0025] The following description sets forth exemplary aspects of the present disclosure. It should be recognized, however, that such description is not intended as a limitation on the scope of the present disclosure. Rather, the description also encompasses combinations and modifications to those exemplary aspects described herein.

[0026] The present embodiments comprise a method and a system for privacy preserving Al-based content generation.

[0027] One technical problem dealt with by the disclosed subject matter has to do with enabling Artificial Intelligence (Al) content generation without compromising privacy of users. As Al systems become more sophisticated and prevalent, there is a growing potential for personalized Al to assist individuals in creative and intellectual pursuits.

[0028] However, one may also aim to preserve user privacy. This assistance may involve collaborative networks of autonomous agents working together to generate content based on specific hints or persona data.

[0029] One example of such a network may involve just two components: a first component being used for providing hints about a target persona, and a second component being used for generating relevant (and potentially personalized) content for such a persona.

[0030] The content may include, for example, text, image, video, action, functionality that is performed with respect to the persona, or the like. The content may be influenced by the "campaign goal", e.g., the goal for providing the content, and the specific personality data provided by the first component.

[0031] Another rapidly growing and promising direction in the field of content generation is the development of Edge Al models, which consider the multi-faceted patterns of user behavior. Edge models may be designed to focus on understanding the multi-faceted patterns of user behavior.

[0032] Edge models may utilize behavioral information available on client devices of users, while keeping each user’s private information on the user’s client device. The result of such models may be a digital representation of the user that is confined to the user’s client device, and thus protects user privacy.

[0033] One could suggest encapsulating the characteristics of a user of a client device in a single Al ambassador. [0034] However, fully encapsulating an individual's personality and interests in a single Al ambassador raises privacy issues. As the volume of content being produced worldwide is escalating rapidly, individuals may face difficulties in effectively filtering and curating content that aligns with their interests. Balancing the desire for personalized Al assistance with the need to protect user privacy becomes a complex technical problem that needs to be addressed for the responsible and ethical development of Al content generation systems.

[0035] Among the solutions hitherto explored, are solutions such as GOOGLE's initial proposals for FLoC (Federated Eeaming of Cohorts) and Privacy Sandbox, that aim to solve the issue of personalization while maintaining privacy by grouping users into clusters based on their behavior.

[0036] Such solutions may enable advertisers to target relevant ads to users without being exposed each user’s personal identity or interests. However, such solutions may still pose privacy risks, such as exposing users to fingerprinting or unwanted profiling.

[0037] Additionally, or alternatively, such solutions inherently imply that personalization can only be done at a very coarse level. On the other hand, creating the content on the device may be computationally intensive, and in some cases, may not even be feasible due to computing resource constraints.

[0038] One technical solution described hereinbelow is to generate partially adapted contents for a user, by a remote server in communication with a user’s client device, based on partially identifiable descriptors (i.e. characteristics) of the user generated by the user’s client device. A more personalized content may be determined on the user’s client device, based on the partially adapted content.

[0039] In some cases, the personalized content may be selected from the partially adapted content, based on the respective partially identifiable descriptors and their relation to the user’s actual data (with or without taking into consideration the relevant domain of the content).

[0040] In other cases, the personalized content may be generated based on the partially adapted content and based on the relation between the respective descriptors upon which the partially adapted content was created and a complete descriptor of the user.

[0041] With such a solution, identifiable data of the user, such as PII, may not be provided to the external servers in charge of creating the content. Instead, different combinations of potential PIIs, some of which are actually reflecting the user, and some are not (say false PIIs), may be divulged to the external servers, thereby preserving the user’s privacy. While the expensive creation of content is performed on external servers, only recombination thereof is required on the client device side.

[0042] In some exemplary embodiments, a personality profile of each user may be derived through on-device inference using behavioral models. The profile may be securely retained on the user's own device. A digital representation of the user’s personality may be utilized as the foundation for the creation of customized descriptors (also referred to as Al ambassadors).

[0043] This process may employ algorithmic techniques such as Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), Diffusion Models, or the like.

[0044] In some cases, a cryptographically secure random key cipher, also confined to the user's device, may be utilized to generate partially identifiable descriptors that are unique and personalized.

[0045] The partially identifiable descriptors or Al ambassadors may be “digital cousins” that act as digital extensions of the user, but for privacy reasons, none of them exactly mirrors the user's personality. Some partially identifiable descriptors may be entirely "fake" or "misleading", having little in common with the actual user. This intentional discrepancy may help ensure that when these partially identifiable descriptors operate outside of the device, they can engage with content networks effectively without revealing the user's true identity or personal details.

[0046] In some exemplary embodiments, the partially identifiable descriptors may participate in creative networks and marketplaces for content creation, reacting based on their individual Al models. The content produced for a specific partially identifiable descriptor, or a subset of partially identifiable descriptors, may be tailored to the unique personal characteristics and interests embodied by those partially identifiable descriptors.

[0047] In some exemplary embodiments, the user’s client device may receive the content generated for each partially identifiable descriptor associated with the given user.

[0048] The final personalized content for the user may then be reassembled from the received set of outputs. In some cases, the reassembling may be performed using a local algorithm that applies the private key cipher on the content based on the partially identifiable descriptors.

[0049] One technical effect of utilizing the disclosed subject matter may be to enable personalization and contextualization of content while preserving privacy, without compromising the arbitrary fine granularity of personalization, and without exposing users to fingerprinting or unwanted profiling.

[0050] Disclosed embodiments thus aim to preserve user privacy in the context of Al content generation compared to a single fully representative Al ambassador. The proposed systems and methods may preserve user privacy by utilizing a distributed, decentralized, and possibly, intentionally misleading Al ambassadors, combined with on-device processing and cryptographic techniques, to enable Al content generation while prioritizing and preserving user privacy.

[0051] Another technical effect of utilizing the disclosed subject matter is to enable personalized filtering of high volumes of generated content.

[0052] A yet another technical effect of utilizing the disclosed subject matter is to provide a framework for secure human collaboration with creative Al.

[0053] A yet another technical effect of utilizing the disclosed subject matter is to facilitate the monetization of the final content for both the vendor and the end user, without compromising user privacy, while maintaining all user private information on the user’s device and without divulging such information to external devices or servers.

[0054] The disclosed subject matter may provide for one or more technical improvements over any pre-existing technique and any technique that has previously become routine or conventional in the art. Additional technical problem, solution and effects may be apparent to a person of ordinary skill in the art in view of the present disclosure.

[0055] The principles and operation of a system and method according to the disclosed subject matter may be better understood with reference to the drawings and accompanying description. The disclosed subject matter is capable of other embodiments or of being practiced or carried out in various ways. Also, it is to be understood that the phraseology and terminology employed hereinbelow is for the purpose of description only and should not be regarded as limiting.

[0056] Reference is now made to Fig. 1 which is a simplified flowchart illustrating a first exemplary method of content generation, according to an exemplary embodiment of the disclosed subject matter. [0057] The exemplary method includes steps that at least one computer processor, say a computer processor that is a part of a circuit (i.e. hardware and associated circuitry) of a computer, is programmed to perform.

[0058] Optionally, the computer is a user’s client device that is in communication with a computer that is used for generating Artificial Intelligence (Al) based content, say with a computer that provides content to a plurality of client devices, say in a creative network.

[0059] In the first exemplary method, there is extracted 101 a set of characteristics of a user of the client device from data retained on the client device, say from data retained in one or more database(s) and/or in file(s) that are stored on the client device. The characteristics may additionally or alternatively, be extracted from data extracted from readings of one or more sensor(s) of the client device.

[0060] Next in the method, a respective subset of the extracted set of characteristics of the user is encapsulated 102 in each Al ambassador of a group of one or more Artificial Intelligence (Al) ambassador(s). The encapsulation 102 may include storing a representation of the subset of characteristics in a data structure that is a part of the Al ambassador, the provision of a mechanism (say a function) for reading the characteristics stored on the Al ambassador, etc., as known in the art.

[0061] Thus, in one example, the extracted 101 characteristics include a user’ s age, gender, favorite soccer club (as extracted 101 from the user’s web browsing history), driving habits (say an average time of driving per day as calculated based on location data extracted from the user’s client device’s GPS receiver), user’s most frequently played songs (based on the user’s playlists retained on YouYube™ or another media content application), etc.

[0062] In one case, in the example, the user’s most frequently played songs are encapsulated 102 on a first Al ambassador, whereas the user’s driving habits are encapsulated 102 in a second Al ambassador, etc.

[0063] Next, one or more Al ambassador(s) of the group are used by the client device, for receiving 103 content that is selected by the content generating computer. The content is selected for the Al ambassador (and possibly, also generated) by the content generating computer based on the characteristics encapsulated in the Al ambassador. The characteristics encapsulated in the Al ambassador may be accessed by the content generating computer. As an example an Al model may be utilize the characteristics encapsulated in the Al ambassador to generate a personalize content that is relevant to the characteristics of the user, as represented by the Al ambassador.

[0064] Thus, in one exemplary case, only the first Al ambassador is used for the receiving 103, and the received 103 content includes songs selected by the content generating computer based on the user’s most frequently played songs as encapsulated 102 in the first Al ambassador.

[0065] In a second exemplary case, only the second Al ambassador is used for the receiving 103, and the received 103 content includes podcasts selected by the content generating computer based on the user’s driving habits as encapsulated 102 in the second Al ambassador. The driving habits may include, for example, the user’s daily duration and time of travel, such that the podcasts are selected according to user’s average length of drive and the drive’s being in the morning or rather, in the evening.

[0066] In a third exemplary case, a Large Language Model (LLM), or another Al model, may be utilized to generate the content. The LLM model may be provided with the characteristics of the user, as depicted in the first and second Al ambassadors, so as to bias the model to generate content that is relevant to those characteristics. For example, a generated image may include a person. Based on the location of the user being within China, which may depicted in an Al ambassador, the person may be generated so as to appear Chinese. Such a personalization may be utilized to provide overall look that is more relevant to the user, based on his regular environment.

[0067] Then, the received 103 contents may be presented 104 to the user, say by playing a video on the client device’s screen or by presenting an article on the device’s screen.

[0068] Optionally, the encapsulating 102 further comprises encapsulating 102 one or more false characteristic(s) of the user in at least one of the Al ambassadors of the group, thus potentially defeating attempts to use the encapsulated 102 characteristics, for exposing the user’s identity or otherwise compromising the user’s privacy.

[0069] Optionally, the method further comprises selecting the part of the received 103 content for the presenting 104, say using one or more of the extracted 101 characteristics of the user that are not encapsulated 102 in the Al ambassador used for receiving 103 the content. As a result, content presentation 104 to the user becomes more finely targeted, without compromising user privacy. [0070] Thus, in one example, the podcasts received 103 using the second Al ambassador, that are not appropriate to younger ages, are filtered out based on the age of the user, that is not encapsulated 102 in the second Al ambassador, but is still retained on the client device.

[0071] In a second example, the podcasts 103 are selected according to an age group (say 3-18) that is represented in the characteristics encapsulated 102 in the Al ambassador. However, podcasts that are meant to be presented to children of younger ages (say 3-4 years old) are filtered out based on the user’s accurate age (say 14 years old) that is retained on the user’s client device (but is not encapsulated in the Al ambassador).

[0072] Optionally, the data retained on the client device comprises data extracted from a reading of at least one sensor of the client device, say data extracted from a reading of at least one sensor of the client device.

[0073] The at least one sensor may include, for example, a gyroscope, a Wi-Fi receiver, a blue tooth receiver, a charger connection sensor, an accelerometer, a GPS receiver, a camera, a magnetometer, a proximity sensor, an ambient light sensor, a microphone, a touchscreen sensor, a fingerprint sensor, a pedometer, a barometer, a thermometer, and an air humidity sensor.

[0074] Optionally, the method further comprises consolidating content received 103 from the content generating computer using at least two Al ambassadors of the group, in a single format used in the presenting 104, say in a single webpage, document (say a pdf document), etc.

[0075] Referring now to Fig. 2, which illustrates a flowchart depicting a method for privacy-preserving content generation using Al ambassadors and creative networks, according to various embodiments.

[0076] The method begins with a step 210 of modeling a personality representation of a user on a client device. In some cases, the Personality Modeling Module 421 may perform this modeling based on behavioral data and other characteristics extracted from data stored locally on the client device. The personality representation may include various identifiable information about the user, such as demographic data, preferences, and behavioral patterns.

[0077] In a step 220, the method proceeds to generate and deploy Al ambassadors into creative networks. The Algorithmic Ambassador Generation System 422 may create multiple distinct Al ambassadors, each representing different aspects of the user's modeled personality. In some cases, the group of Al ambassadors may comprise at least two Al ambassadors, providing a diverse set of personas to engage with content generation platforms.

[0078] Importantly, during the generation of Al ambassadors, the method may include encapsulating at least one false characteristic of the user in at least one of the Al ambassadors. This intentional inclusion of misleading information may help to preserve user privacy by preventing any single Al ambassador from fully representing the user's true identity.

[0079] The method continues with a step 230, where the Al ambassadors autonomously participate in personality-conditioned generation and monetization of content. In this step, each Al ambassador may independently contribute to content creation based on its unique subset of user characteristics, ensuring varied and personalized content output.

[0080] In a step 240, the method involves aggregating the outputs of the Al ambassadors. The Content Recombination Module 441 may perform this aggregation, collecting diverse content generated for multiple Al ambassadors.

[0081] Finally, in a step 250, the method applies a locally stored key cipher to filter and recombine the content. The Private Key Cipher 431 may be used in this step to enable the client device to process the generated content, filtering out irrelevant items and recombining the remaining content into a consolidated personalized content 480 tailored to the user's preferences.

[0082] This method may allow for personalized engagement with creative Al while prioritizing user privacy and security by distributing content creation across multiple partial representations of the user and employing on-device filtering and recombination.

[0083] The method of Fig. 2 includes steps that may be performed by at least one computer processor, such as a computer processor that is part of a circuit (i.e. hardware and associated circuitry) of a computer.

[0084] In some aspects, the computer may be a user's client device that communicates with a computer used for generating Artificial Intelligence (Al) based content, for example a computer that provides content to multiple client devices connected in a creative network, as described in further detail above.

[0085] The method of Fig. 2 may be utilized for privacy-preserving distributed content generation using algorithmically generated Al ambassadors. [0086] The method of Fig. 2 aims to balance the generation of personalized content with user privacy considerations, addressing challenges associated with increasingly sophisticated artificial intelligence (Al) systems and the need for privacy-preserving content creation.

[0087] In Step 210 of the method, a personality representation of the user may be modeled on a user's client device based on the user's behavioral data and/or other user characteristics 101 extracted on the client device.

[0088] The personality representation of the user may comprise identifiable information of the user, such as gender, age, user preferences and behavioral patterns such as web browsing habits, activity on the client device, GPS-based geographical location, etc.

[0089] The modeling 210 of the user's personality representation may involve capturing explicit user preferences, implicit user preferences, behavioral models of user, etc.

[0090] In some cases, some information used for the modeling 210 may be determined using an Artificial Intelligence (Al) model, a Machine Learning (ML) model, a Deep Learning (DL) model, a classifier, a predictor, or the like, for predicting an identifiable information datum about the user.

[0091] For example, an Al model may be utilized to determine, based on available information about the user, including private information that may be located locally on the device, whether the user can be categorized as a "VIP client", "premium potential", "high affluence", "chum risk", etc.

[0092] The determination may be based on whether or not the user is categorized according to a desired model-determined category. In some aspects, the modelling of the personality representation of the user may be performed by on-device inference without transmitting sensitive data over external networks to help ensure that sensitive information remains securely stored on the user's device, preserving user privacy.

[0093] In Step 220 of the method of Fig. 2, one or more Al ambassadors may be generated and used for encapsulating a part of the user characteristics that are extracted on the user's client device.

[0094] The Al ambassadors may be deployed into an Al-based creative network and/or otherwise used for receiving content (such as advertisement videos) from one or more computers that generate content based on Al. [0095] In some cases, each of the Al ambassadors may represent the user in one or more aspects and participate in content creation within decentralized networks. The set of Al ambassadors may thus provide a diverse set of personas to engage with Al-based content generating computers.

[0096] For example, each Al ambassador may represent the user only in a portion of the user's characteristics extracted on the user's client device.

[0097] Thus, in one example, a first Al ambassador may hold (i.e. encapsulate) only demographic information (such as age group and gender) on the user, whereas a second Al ambassador may hold only characteristics related to a certain behavioral aspect (e.g., sports preferences) of the user.

[0098] In some cases, each Al ambassador may be true to the user's persona with relation to a different aspect, such that different Al ambassadors may be useful in different cases, depending on the relevant topic of the content to be received.

[0099] In some aspects, a static set of Al ambassadors may be created for each user's client device in advance, such that the ambassadors are created once. The Al ambassadors may then be utilized by different content generators that generate content based on topics, such that the generated content is received from a computer that implements the respective generator.

[0100] Additionally, or alternatively, different sets of Al ambassadors may be created on the fly and on demand. In some cases, the different Al ambassadors may be generated without depending on their expected use.

[0101] In other cases, the Al ambassadors may be generated while taking into account the topics that are relevant to the content creation. For example, there may be a different set of Al ambassadors created for selecting news topics, for selecting an advertisement about a car, for deciding to update a CRM with an indication of churn likelihood of the user, or the like.

[0102] In some aspects, the set of Al ambassadors may comprise about 100 Al ambassadors, about 50 Al ambassadors, or a similar number. Each Al ambassador may partially represent the user.

[0103] In some cases, the Al ambassador may be constrained to be similar to the persona within a predetermined similarity measurement. For example, 30% of the information may be identical to the persona information. As another example, a distance measurement may be computed, and a maximal distance may be defined. [0104] In some aspects, none of the Al ambassadors may be 100% representative of the user, thereby helping to keep the user un-identifiable by external devices, and preventing a singling out of the user.

[0105] As an example, a set of 100 Al ambassadors may be created in Step 220, based on the personality representation of the user. Each of the 100 Al ambassadors may represent the user in at least 60% of their personality representation, for instance by being used to encapsulate only 60% of the user's characteristics extracted on the user's client device. The additional 40% representation may comprise characteristics that do not accurately reflect the user (i.e. inaccurate information, misleading information, or the like).

[0106] In some aspects, each Al ambassador may comprise different types of user characteristics, such as demographic information, pictures, browsing data, or the like.

[0107] In Step 230 of the method of Fig. 2, the Al ambassadors may autonomously participate in personality-conditioned generation and monetization of the content.In some aspects, creative content may be configured to generate personality-conditioned content based on the user's characteristics encapsulated in each respective Al ambassador.

[0108] Each Al ambassador may thus independently contribute to content creation based on its unique personality representation (i.e. characteristics encapsulated therein), helping to ensure a varied and personalized content output aligned with user preferences.

[0109] The content received based on the Al ambassadors may comprise different types of Al-generated and/or Al-selected content that is likely to be relevant to the user, such as text, images, articles, campaigns, or the like.

[0110] The received content may be associated with one or more topics, such as campaigns for certain merchandise, articles of certain disciplines, etc..

[0111] In one example, sport content that is created for a first Al ambassador may comprise sport activity related features, whereas news content created for a second Al ambassador may comprise features associated with news consumption features, etc.

[0112] It may be noted that the entity that creates the content (e.g., the creative network) may not be aware of which portion of the Al ambassador is accurate and which portion is inaccurate, but may still generate and/or select the personalized content for a persona description aspect represented by the Al ambassador. [0113] In Step 240 of the method, receiving the content using the Al ambassadors may be followed by aggregating and consolidating the received content by a content aggregation system, and then presenting the consolidated content to the user.

[0114] In some aspects, such aggregation may comprise generating, based on a collection of diverse content generated for the multiple Al ambassadors, personalized content adapted to the user, based on a true identifier thereof.

[0115] In some aspects, the aggregation may involve a selection of the most suitable content for the persona, for example by filtering out parts of the content that do not fit user characteristics that are retained on the client device, but are not encapsulated in the Al ambassadors (for instance, for privacy protecting purposes).

[0116] In some cases, the selection may be performed on the user's client device, by determining which of potential content items is best suited for the persona.

[0117] It is noted that the computation complexity of determining which content, out of a small set of potential contents (e.g., 100 contents) is best suited for the persona may be significantly smaller than matching content to the persona from a large set of potential content items, as is done in the creative network. Hence, the local selection in the client device may be computationally feasible.

[0118] Additionally, or alternatively, another form of local selection may be employed on the user's client device.

[0119] The selection may be made based on a determination of the most suitable Al ambassador for the content.

[0120] For example, an Al ambassador which best matches the persona with respect to a specific aspect relating to the topic of the content may be determined and the content matched thereto may be selected.

[0121] In some cases, a projection of properties relating to the topic may be utilized and a distance between the projections of the Al ambassadors and the persona may be computed to select the closest Al ambassador relating to the topic.

[0122] The received content is presented to the user.

[0123] Additionally, or alternatively, the content that is generated and received using two or more Al ambassadors, such as two or more content items, may be consolidated prior to presentation to the user. [0124] Thus, in some aspects, the content may be consolidated based on two or more content items generated for respective two or more Al ambassadors, based on the representativity of user persona within each respective one of the Al ambassadors.

[0125] As an example, consider content creation that relates to a page with articles. The page that is created for the persona may be a consolidation of several articles received using different Al ambassadors, such that mix-and-matching between the articles appear in different created pages.

[0126] For example, the sports section may be created by including articles created for Al ambassadors that accurately reflect the persona with respect to sports and news articles created for different Al ambassadors that accurately reflect the persona with respect to news.

[0127] In Step 250 of the method of Fig. 2, the content may be subjected to ciphering and filtering, by applying locally (i.e. on the client device) stored key cipher to filter and recombine the content.

[0128] In some aspects, applying the local key cipher may enable the user's device to filter out irrelevant content, recombining content associated with different Al ambassadors into personalized content tailored to the user's preferences. This may help ensure that only the user's device possesses the capability to decrypt and process the generated content.

[0129] By distributing content creation and monetization across Al ambassadors while filtering via an on-device cipher, personalized engagement with creative Al may be allowed while prioritizing user privacy and security.

[0130] Reference is now made to Fig. 3, which illustrates a block diagram of a system 3000 for privacy-preserving content generation, according to an exemplary embodiment of the disclosed subject matter.

[0131] The system 3000 may be implemented using electric circuits, computer software, computer hardware, or any combination thereof.

[0132] In some aspects, the system 3000 may be implemented on a user's client device that communicates with one or more computers 3001 used as Artificial Intelligence (Al) based content generating computer(s), for example to implement a creative network.

[0133] The system 3000 includes a computer processor 301 and a computer memory 302. The computer memory 302 may comprise, for example, a CD-ROM, a Hard Disk Drive (HDD), a Solid-State Drive (SSD), a memory component of a computer processor, or other storage media.

[0134] The computer memory 302 may store instructions that are executable by the computer processor 301, other parts of the circuitry that makes up the system 3000, or both, to perform steps of the methods described and illustrated using Fig. 1-2 hereinabove.

[0135] The instructions may include steps for extracting a set of characteristics of a user of the client device from data retained on the client device. This data may be stored in one or more database(s) and/or file(s) on the client device.

[0136] In some cases, the characteristics may be extracted from readings of one or more sensor(s) of the client device.

[0137] The instructions may further include steps for encapsulating a respective subset of the extracted set of characteristics of the user in each Al ambassador of a group of one or more Al ambassador(s)).

[0138] For example, the extracted characteristics may include a user's gender, web browsing habits, driving habits, and other relevant data.

[0139] In one implementation, the user's most frequently played songs may be encapsulated in a first Al ambassador, while the user's driving habits may be encapsulated in a second Al ambassador.

[0140] The instructions may further include steps for using one or more Al ambassador(s) of the group to receive content that is selected and/or generated by the content generating computer.

[0141] The content may be selected for the Al ambassador by the content generating computer based on the characteristics encapsulated in the Al ambassador, as described in further detail and illustrated using Fig 1-2 hereinabove.

[0142] In one example, only a first Al ambassador may be used by the client agent for receiving the content. The received content may include songs selected by the content generating computer based on the user's most frequently played songs and/or a characteristic thereof (e.g. country music) as encapsulated in the first Al ambassador.

[0143] In another example, only the second Al ambassador may be used by the client agent for receiving the content. The received content may include podcasts selected by the content generating computer based on the user's driving habits as encapsulated in the second Al ambassador.

[0144] The instructions may further include steps for presenting the content or a part thereof to the user, for example by playing a video or a clip on the client device's screen.

[0145] In some aspects, the steps of encapsulating may further comprise steps for encapsulating one or more false characteristic of the user in at least one of the Al ambassadors of the group. This may help protect the user's privacy by potentially defeating attempts to use the encapsulated characteristics to expose the user's identity.

[0146] The instructions may further include steps for selecting the part of the received content for presentation, potentially using one or more of the extracted characteristics of the user that are not encapsulated in the Al ambassador used for receiving the content.

[0147] For instance, podcasts received using the second Al ambassador that are not appropriate for younger ages may be filtered out based on the age of the user, which may be retained on the client device but not encapsulated in the second Al ambassador.

[0148] In some implementations, the data retained on the client device may comprise data extracted from a reading of at least one sensor of the client device.

[0149] The sensors may include, for example, a gyroscope, a Wi-Fi receiver, a Bluetooth receiver, a charger connection sensor, an accelerometer, a GPS receiver, a camera, a magnetometer, a proximity sensor, an ambient light sensor, a microphone, a touchscreen sensor, a fingerprint sensor, a pedometer, a barometer, a thermometer, and an air humidity sensor.

[0150] The instructions may further include steps of consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single format used in the presenting. This format may be a single webpage, document (e.g. a PDF document), podcast, playlist, or other suitable format.

[0151] In some implementations, the instructions may be used to model a personality representation of the user from the user's behavioral data extracted on the client device, potentially using Al methods.

[0152] The personality representation of the user may include identifiable information such as gender, age, web browsing habits, activity on the client device, GPS -based geographical location, and other relevant data. [0153] The modeling may involve capturing explicit user preferences, implicit user preferences, behavioral models of user activity, or similar data.

[0154] In some aspects, the Al ambassadors may be provided to creative networks and/or used for receiving content (such as advertisement videos) from one or more computers that generate content based on Al.

[0155] Each Al ambassador may represent the user in a different way and participate in content creation within decentralized networks. The set of Al ambassadors may thus provide a diverse set of personas to engage with Al-based content generating computers.

[0156] For example, each Al ambassador may represent the user only in a portion of the user's characteristics extracted on the user's client device.

[0157] In one implementation, a first Al ambassador may hold only demographic information (e.g., age and gender) on the user, while a second Al ambassador may hold only characteristics related to a certain behavioral aspect (e.g., sports preferences) of the user.

[0158] In some cases, each Al ambassador may be true to the user's persona with relation to a different facet, such that different Al ambassadors may be useful in different scenarios, depending on the relevant topic of the content to be received.

[0159] In some implementations, each Al ambassador may comprise different types of user characteristics, such as demographic information, pictures, browsing data, or other relevant data.

[0160] In some cases, one or more of the Al ambassadors may be used to encapsulate true characteristics of the user along with false characteristics, potentially helping to protect the user's privacy.

[0161] In some implementations, the system 3000 may be designed to be scalable, allowing for the addition of more computers 3001 or expansion of processing capabilities as needed. This scalability may enable the system to handle increasing amounts of data or users while maintaining performance and privacy protections.

[0162] Reference is now made to Fig. 4, which illustrates a block diagram of a second system for content generation, according to an exemplary embodiment of the disclosed subject matter.

[0163] In some implementations, a system 4000 for content generation may comprise an Al based Creative Network 4100. [0164] The Al based Creative Network 4100 may include one or more Al based content generating computers that interact with one or more Al Ambassador(s) 402 to perform content generation and/or selection for the Al ambassadors 402, using a content generation agent 401.

[0165] The Creative Network 4100 may be implemented on a server computer and/or other computers used for content generation.

[0166] The Creative Network 4100 may be configured to generate personalized content based on one or more Al ambassadors 402, in which descriptors (i.e. characteristics) of the user obtained from the user's client device are encapsulated.

[0167] In some aspects, the Creative Network 4100 may be configured to generate and/or select content based on the Al ambassadors 402 using Al techniques, potentially incorporating content from third party platforms (e.g., publishers or marketplaces).

[0168] The Creative Network 4100 may provide multiple pieces of content to the user's client device 4200 based on the Al ambassador(s) 401 that expose their respective subsets of user characteristics to the Creative Network 4100.

[0169] The client device 4200 may determine which of the content generated by the Creative Network 4100, or which combination thereof, should be presented to the client device's 4200 user.

[0170] In some implementations, the client device 4200 may include a Personality Modeling Module 421.

[0171] The Personality Modeling Module 421 may be configured to capture a digital representation 423 of the user's personality from explicit user preferences and behavioral models of user activity, thus extracting one or more of the user's characteristics.

[0172] The Personality Modeling Module 421 may provide the digital representation 423 of the user's persona to an Algorithmic Ambassador Generation System 422 implemented on the client device 4200.

[0173] The Algorithmic Ambassador Generation System 422 may be configured to create multiple distinct Al ambassadors 402, such as Al ambassador 1, Al ambassador 2, and Al ambassador 3 shown in Fig. 4.

[0174] In some implementations, each Al ambassador 402 may encapsulate a different subset of characteristics from the set of user characteristics that make up the locally stored personality representation of user 423. These characteristics may be extracted from data retained locally on the client device, such as from readings of sensors on the client device.

[0175] The Algorithmic Ambassador Generation System 422 may apply generative Al techniques to generate different descriptors (i.e. characteristics) of the user based on the digital representation of persona generated by the Personality Modeling Module 421.

[0176] In some aspects, each Al ambassador 402 may represent the user partially, for example by comprising incomplete portions of the digital representation of persona 423, or by comprising modifications of the digital representation of persona 423. This approach may help ensure that no single descriptor fully identifies the user.

[0177] As an example, the digital representation 423 of persona may be associated with a "female user, aged 25-35, located in NYC, USA, commuting from home to work, and performing grocery shopping between 14:00-15:00".

[0178] In this example, Al Ambassador 1 may encapsulate the user characteristics of: "female user, aged 25-35, located in Haifa, Israel", Al Ambassador 2 may encapsulate: "user aged 25-35, located in NYC, USA, commuting from home to work" and Al Ambassador 3 may encapsulate: "user located in NYC, USA, performing grocery shopping between 14:00-15:00".

[0179] The Creative Network 4100 may be configured to generate respective content 461 for each of the Al Ambassadors 402, in accordance with an entity interested in distributing the content to users. This entity may include, for example, an advertiser, a campaign provider, or another content provider.

[0180] Additionally, the Creative Network 4100 may generate the respective content 461 for each Al Ambassador 402 based on a digital representation of an aspect of the user's persona 43 derived from the user characteristics encapsulated in the Al Ambassador 402.

[0181] In some implementations, the portion of data representing the user in each Al ambassador 402 may be known to the user's client device 4200, but may not be provided to the Creative Network 4100 or any other external device or platform.

[0182] Additionally, or alternatively, a Private Key Cipher 431 may be cryptographically generated on the client device 4200 to connect each Al ambassador 402 to the digital representation of persona 423 of the user, for example by indicating which portions truly represent the user. [0183] The Private Key Cipher 431 may also be used for the finalization of personalized content 461 creation and filtering by a Content Recombination Algorithm 441 implemented as a module on the client device.

[0184] The Content Recombination Algorithm 441 may determine the relevance of the content 461 provided by Creative Network 4100, or the relevance of specific portions thereof to the user, based on information included in the Private Key Cipher 431.

[0185] The Content Recombination Algorithm 441 may also perform content aggregation and consolidation of the content 461 provided by Creative Network 4100 in a manner that matches the digital representation of persona 423 of the user, thereby creating consolidated personalized content 480 to be presented to the user.

[0186] Referring now to Fig. 5, which illustrates a block diagram of a privacy-preserving content generation system, according to various embodiments. The system includes a client device 500 that communicates with a network system 540. The client device 500 contains several components that work together to generate and process content while maintaining user privacy.

[0187] Within the client device 500, a cipher module 510 and a modeling module 520 provide inputs to a generation system 530. The modeling module 520 may process user characteristics and behavioral data, while the cipher module 510 may manage encryption aspects of the system. In some cases, the modeling module 520 may extract data from readings of at least one sensor of the client device 500. The at least one sensor may comprise at least one of: a gyroscope, an accelerometer, a GPS receiver, a camera, a microphone, or a touchscreen sensor. The modeling module 520 may utilize machine learning algorithms to analyze the sensor data and derive user characteristics.

[0188] The generation system 530 may create multiple Al ambassadors, represented by Al ambassador 551, Al ambassador 552, and Al ambassador 550. Al ambassadors 551 and 552 may operate independently within the client device 500, while Al ambassador 550 may interface with the network system 540. In some cases, the generation system 530 may encapsulate at least one false characteristic of the user in at least one of the Al ambassadors. The generation system 530 may employ advanced techniques such as federated learning or differential privacy to create diverse yet privacy-preserving Al ambassadors.

[0189] The network system 540 includes a creation agent 560 that may generate content based on information received from Al ambassador 550. The creation agent 560 may utilize natural language processing and generative Al models to produce personalized content. The generated content may flow back through the system to a recombination module 570 within the client device 500. In some implementations, the network system 540 may incorporate a distributed ledger or blockchain technology to ensure transparency and security in content generation and transmission.

[0190] The recombination module 570 may receive inputs from both the cipher module 510 and the Al ambassadors (551, 552). In some cases, the recombination module 570 may select a portion of the received content for presenting to the user based on at least one extracted characteristic not encapsulated in the Al ambassador used for receiving the content. The recombination module 570 may employ advanced filtering algorithms and user preference models to tailor the final content presentation.

[0191] The recombination module 570 may consolidate content received from the network system 540 using at least two Al ambassadors of the group (550, 551, 552) into a single format for presenting to the user. The single format may comprise a single webpage or a single document. In some implementations, the recombination module 570 may utilize adaptive layout algorithms to optimize the presentation of consolidated content across different device types and screen sizes.

[0192] The dashed lines in the diagram indicate data flow paths between the various components, showing how information may move through the system while maintaining separation between the client device 500 and network system 540. This separation may be further enhanced by implementing secure communication protocols and end-to-end encryption between components.

[0193] The dashed lines in the diagram indicate data flow paths between the various components, showing how information may move through the system while maintaining separation between the client device 500 and network system 540.

[0194] A number of implementations have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the disclosure. Accordingly, other implementations are within the scope of the following claims.

Claims

1. A method of privacy-preserving content generation, comprising: extracting a set of characteristics of a user from data retained on a client device; encapsulating subsets of the extracted characteristics in a plurality of Artificial Intelligence (Al) ambassadors; using at least one of the Al ambassadors to receive content selected by a remote content generating computer based on the encapsulated characteristics; and presenting at least a portion of the received content to the user.

2. The method of Claim 1, wherein encapsulating subsets of the extracted characteristics comprises encapsulating at least one false characteristic of the user in at least one of the Al ambassadors.

3. The method of Claim 1, further comprising selecting the portion of the received content for presenting to the user based on at least one extracted characteristic not encapsulated in the Al ambassador used for receiving the content.

4. The method of Claim 1, wherein the data retained on the client device comprises data extracted from readings of at least one sensor of the client device.

5. The method of Claim 4, wherein the at least one sensor comprises at least one of: a gyroscope, an accelerometer, a Global Positioning System (GPS) receiver, a camera, a microphone, or a touchscreen sensor.

6. The method of Claim 1, further comprising consolidating content received from the content generating computer using at least two Al ambassadors of the plurality of Al ambassadors into a single format for presenting to the user.

7. The method of Claim 6, wherein the single format comprises a single webpage or a single document.

8. A system for privacy-preserving content generation, comprising: a client device comprising a processor and a memory storing instructions that, when executed by the processor, cause the client device to: extract a set of characteristics of a user from locally retained data; generate a plurality of Artificial Intelligence (Al) ambassadors, each encapsulating a subset of the extracted characteristics; communicate with a remote content generating computer using at least one of the Al ambassadors to receive content selected based on the encapsulated characteristics; and present at least a portion of the received content to the user.

9. The system of Claim 8, wherein at least one of the Al ambassadors encapsulates at least one false characteristic of the user.

10. The system of Claim 8, wherein the instructions further cause the client device to select the portion of the received content for presenting to the user based on at least one extracted characteristic not encapsulated in the Al ambassador used for receiving the content.

11. The system of Claim 8, wherein the locally retained data comprises data extracted from readings of at least one sensor of the client device.

12. The system of Claim 8, wherein the instructions further cause the client device to consolidate content received from the content generating computer using at least two Al ambassadors of the plurality of Al ambassadors into a single format for presenting to the user.

13. The system of Claim 12, wherein the single format comprises a single webpage or a single document.

14. A non-transitory computer-readable medium storing instructions that, when executed by a processor of a client device, cause the client device to perform a method of privacy-preserving content generation, the method comprising: modeling a personality representation of a user based on behavioral data stored on the client device; creating a plurality of Artificial Intelligence (Al) ambassadors, each representing a partial aspect of the modeled personality representation; providing the Al ambassadors to a remote creative network for content generation, wherein the remote create network is configured to utilize the Al ambassadors or portion thereof to generate content; and aggregating content received from the remote creative network to produce personalized content for the user.

15. The non-transitory computer-readable medium of Claim 14, wherein creating the plurality of Al ambassadors comprises encapsulating at least one false characteristic of the user in at least one of the Al ambassadors.

16. The non-transitory computer-readable medium of Claim 14, wherein the behavioral data comprises data extracted from readings of at least one sensor of the client device.

17. The non-transitory computer-readable medium of Claim 14, wherein aggregating the content received from the remote creative network comprises applying a local key cipher to filter and recombine the content.

18. The non-transitory computer-readable medium of Claim 17, wherein applying the local key cipher comprises selecting content based on a determination of which Al ambassador best matches the modeled personality representation with respect to a specific aspect relating to a topic of the content.

19. A method of content generation, the method comprising steps executed by a client device in communication with an Artificial Intelligence (Al) based content generating computer, the steps comprising: extracting a set of characteristics of a user of the client device from data retained on the client device; encapsulating a respective subset of the extracted set of characteristics of the user in each ambassador of a group of at least one Al ambassador; using at least one Al ambassador of the group, for receiving content selected by the content generating computer, the content being selected by the content generating computer based on the characteristics encapsulated in the Al ambassador; and presenting at least a part of the received content selected by the content generating computer, to the user.

20. The method of Claim 19, wherein the group comprises at least two Al ambassadors.

21. The method of Claim 19, wherein said encapsulating further comprises encapsulating at least one false characteristic of the user in at least one of the Al ambassadors of the group.

22. The method of Claim 19, further comprising selecting the part of the received content for said presenting.

23. The method of Claim 19, further comprising selecting the part of the received content for said presenting using at least one of the extracted characteristics of the user not encapsulated in the Al ambassador used for obtaining the content.

24. The method of Claim 19, wherein the data retained on the client device comprises data extracted from a reading of at least one sensor of the client device.

25. The method of Claim 19, further comprising extracting the data retained on the client device from a reading of at least one sensor of the client device.

26. The method of Claim 19, wherein the data stored on the client device comprises data extracted from a reading of at least one sensor of the client device, the at least one sensor comprising at least one of the group consisting of: a gyroscope, a Wi-Fi receiver, a blue tooth receiver, a charger connection sensor, an accelerometer, a GPS receiver, a camera, a magnetometer, a proximity sensor, an ambient light sensor, a microphone, a touchscreen sensor, a fingerprint sensor, a pedometer, a barometer, a thermometer, and an air humidity sensor.

27. The method of Claim 19, further comprising consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single format used in said presenting.

28. The method of Claim 19, further comprising consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single webpage used in said presenting.

29. The method of Claim 19, further comprising consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single document used in said presenting.

30. A system for data distribution in a computer network, the system comprising a circuit comprising a computer processor and a computer memory storing instructions that are executable by the computer processor, for performing the steps of: extracting a set of characteristics of a user of the client device from data retained on the client device; encapsulating a respective subset of the extracted set of characteristics of the user in each ambassador of a group of at least one Artificial Intelligence (Al) ambassador; using at least one ambassador of the group, for receiving content selected by the content generating computer, the content being selected by the content generating computer based on the characteristics encapsulated in the Al ambassador; and presenting at least a part of the received content selected by the content generating computer, to the user.

31. The system of Claim 30, wherein the group comprises at least two Al ambassadors.

32. The system of Claim 30, wherein said encapsulating further comprises encapsulating at least one false characteristic of the user in at least one of the Al ambassadors of the group.

33. The system of Claim 30, wherein the steps further comprise selecting the part of the received content for said presenting.

34. The system of Claim 30, wherein the steps further comprise selecting the part of the received content for said presenting using at least one of the extracted characteristics of the user not encapsulated in the Al ambassador used for receiving the content.

35. The system of Claim 30, wherein the data retained on the client device comprises data extracted from a reading of at least one sensor of the client device.

36. The system of Claim 30, wherein the steps further comprise extracting the data retained on the client device from a reading of at least one sensor of the client device.

37. The system of Claim 30, wherein the steps further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single format used in said presenting.

38. The system of Claim 30, wherein the steps further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single webpage used in said presenting.

39. The system of Claim 30, wherein the steps further comprise consolidating content received from the content generating computer using at least two Al ambassadors of the group, in a single document used in said presenting.