A trusted bottom layer implementation framework and system of a man-machine collaborative intelligent system

Abstract

This invention discloses a trusted underlying implementation framework and system for a human-machine collaborative intelligent system. Through a five-layer closed-loop rigid design, it achieves irreversible hardware-level isolation of raw privacy data, automatic blocking of inducing instructions, full-link traceability of interaction responsibility, non-inducing disclosure of the right to know, positive incentives for compliant behavior, and mandatory binding of trusted authentication with access to national public infrastructure. This invention provides a set of neutral, universal, and unavoidable industry-level security rules, preventing privacy abuse, algorithmic manipulation, responsibility shifting, and human-driven manipulation at the source. While ensuring technological innovation, it provides secure, fair, and sustainable underlying technical support for human-machine collaboration, brain-computer interaction, and hybrid intelligence.

Description

Technical Field

[0001] This invention belongs to the fields of artificial intelligence security, human-computer interaction, data privacy protection and intelligent system governance. Specifically, it relates to a trusted underlying implementation framework for human-computer collaboration, brain-computer interaction and hybrid intelligence, which is applicable to the compliant deployment, secure operation and interconnection of various general artificial intelligence, brain-computer interface devices and embodied intelligence systems. Background Technology

[0002] With the deep development of artificial intelligence and brain-computer interface technology, human-machine collaboration scenarios are becoming increasingly widespread. Intelligent systems with different deployment architectures and interaction modes exhibit significant differences in risk levels, data processing methods, and liability attribution. Existing technologies mostly focus on function implementation and performance optimization, lacking unified, rigid, and implementable underlying trusted standards. This leads to problems in some systems such as blurred data boundaries, privacy leakage risks, decision-making manipulation, unclear liability definition, and imbalanced rights mechanisms.

[0003] Meanwhile, the industry lacks differentiated governance and a robust safety net, easily leading to either a "one-size-fits-all" approach to management or a complete lack of oversight. To achieve a synergistic balance between technological innovation, user rights, public safety, and sustainable industry development, a universal, rigid, and unavoidable trusted underlying framework is urgently needed to provide a standardized and compliant foundational solution for the entire human-machine collaboration industry. Summary of the Invention

[0004] 1. Purpose of the Invention This invention constructs a trusted underlying implementation framework for a human-machine collaborative intelligent system. Through a five-layer closed-loop rigid design, it achieves hardware-level irreversible isolation of original privacy data, full-link traceability of instruction behavior, automatic blocking of induced operations, non-induced disclosure of the right to know, positive binding of rights and compliant behavior, and mandatory linkage between trusted access and national public infrastructure. While ensuring the space for industry innovation, it forms a security bottom line that cannot be bypassed, tampered with, or circumvented.

[0005] 2. System Overall Architecture: A trusted underlying implementation framework for a human-machine collaborative intelligent system, characterized by comprising five indispensable collaborative core modules: 1) The local perception data hardware-level irreversible desensitization and isolation module completes local processing of raw neural signals, EEG signals, perception signals and deep consciousness-related data within a trusted execution environment (TEE). It only outputs mathematically irreversible desensitization intent information, and it is impossible to reconstruct the original consciousness characteristics and decision-making logic through any algorithm, model or data fitting. It implements a hardware one-way output mechanism, and regardless of whether the user authorizes it, the raw data cannot be transmitted out of the local hardware domain. There are no bypassable software switches, backdoors or remote configuration channels.

[0006] 2) The instruction source identification and inducement behavior blocking traceability module adds an unalterable source identifier to each interactive instruction, distinguishing between user-initiated, system-assisted suggestions, and intelligent reasoning results; it automatically identifies, marks, and blocks the execution of inducement, bias, manipulation, and incentive-inducing instructions; the interaction log is stored locally and supports regulatory verification, but does not support deletion, tampering, or overwriting in the cloud; it automatically forms a chain of responsibility, with the service provider bearing primary responsibility for the model-generated content and inducement behaviors.

[0007] 3) User right to know: Non-inducing transparent disclosure modules should disclose the scope of data, permission boundaries, experience differences and risks in a concise, clear and non-technical manner; it is prohibited to induce users to give up privacy protection or reduce security level in exchange for performance improvement, service upgrade and personalization enhancement; ensure the availability of basic services in all modes, and do not build coercive authorization mechanisms.

[0008] 4) The positive matching of rights and interests and the rigid incentive module for compliance establish a positive binding mechanism between the degree of user privacy protection and service quality, personalization adaptation level and resource allocation weight; follow the incentive logic of "high compliance → high adaptation → high stability", prohibit the construction of a business mechanism of "giving up privacy → obtaining stronger performance"; and guide the formation of a safe, sustainable and non-sacrificial human-computer interaction mode.

[0009] 5) Establish a unified trusted authentication system for trusted authentication and national public infrastructure access modules. Only systems that pass the authentication can access national public communication networks, government computing power platforms, public service networks and cross-domain interconnection nodes. Systems that do not meet the minimum trusted requirements of this framework are not qualified to access public infrastructure and cannot achieve large-scale networking and commercial deployment.

[0010] 3. Implementation Method A trusted underlying implementation method for a human-machine collaborative intelligent system, including: - Classifying and identifying the intelligent system according to its deployment method, cloud dependency, and data processing mode; - Achieve irreversible desensitization and physical isolation of original privacy data at the hardware level, and even user authorization cannot transmit the original data; - Mark the source of instructions to identify and block manipulative and suggestive behaviors, and solidify evidence across the entire chain; - Disclose the differences and risks of the model to users in a non-inducing manner, and prohibit authorization through incentives; - Establish a positive and rigid matching mechanism between compliant behavior and service rights; - By forcibly binding trusted authentication with national public networks, computing power, and interconnection infrastructure, the underlying access control of the industry can be achieved.

[0011] 4. Beneficial Effects: 1. Achieve irreversible isolation of privacy data from the hardware level, fundamentally preventing the collection, misuse, and commercialization of consciousness data; 2. Rigidly block manipulative and suggestive interactions to protect users from algorithmic manipulation and exploitation of human weaknesses; 3. Responsibility is traceable and tamper-proof, preventing the transfer and shirking of responsibility from a technical perspective; 4. Guide compliance with positive incentives rather than mandatory constraints, balancing user experience and safety; 5. Access rights are tied to national infrastructure, forming an industry-wide set of rules that no entity can circumvent; 6. The framework is neutral, universal, and non-confrontational, aligns with national governance principles, and is easy to review, adopt, and implement in the industry. Attached Figure Description

[0012] Figure 1 This is a hierarchical topology diagram of the five core modules of the trusted underlying implementation framework of the human-machine collaborative intelligent system of the present invention. It shows the hierarchical relationship of the five core modules and the one-way data transmission path, clarifies the original data isolation boundary of the local hardware domain, and reflects the irreversible design logic of the one-way hardware output.

[0013] Figure 2 This is a flowchart of the steps of the trusted underlying implementation method of the human-machine collaborative intelligent system of the present invention, which fully presents the entire process execution sequence from system classification and identification to binding with the access rights of national public infrastructure.

Claims

1. A trusted underlying implementation framework for a human-machine collaborative intelligent system, characterized in that, include: Local sensing data hardware-level irreversible desensitization and isolation module, Instruction source identification and induced behavior blocking and tracing module User right to know: Non-inducing transparent disclosure module Positive matching of rights and interests and rigid incentive modules for compliance Trusted Authentication and National Public Infrastructure Access Module; The above modules form a closed-loop and indispensable underlying architecture.

2. The frame according to claim 1, characterized in that, The desensitization process is mathematically irreversible, making it impossible to reconstruct the original consciousness, neural signals, and decision-making characteristics in any way. The original data is isolated by the hardware domain and cannot be transmitted externally regardless of user authorization, with no bypass channels available.

3. The frame according to claim 1, characterized in that, The system can automatically identify and block inducing, biased, manipulative, and suggestive instructions and interactions, and solidify the chain of evidence.

4. The frame according to claim 1, characterized in that, It is prohibited to induce users to lower their privacy protection level or abandon data isolation in exchange for performance, experience, or service upgrades.

5. The frame according to claim 1, characterized in that, Trusted authentication is a mandatory prerequisite for accessing national public communication networks, interconnection facilities, and public computing platforms.

6. A trusted underlying implementation method for a human-machine collaborative intelligent system, comprising: System classification and identification; Hardware-level implementation ensures irreversible desensitization and physical isolation of raw data, preventing its external transmission even with user authorization; Mark the source of instructions, block manipulative or misleading behavior, and solidify evidence of responsibility; Disclosure of risks and authority in a non-inducing manner is required, and authorization through inducement of benefits is prohibited. Establish a positive and rigid alignment between compliance and rights; Mandate linking trusted authentication with access rights to national public infrastructure.

7. A computer-readable storage medium or intelligent device storing a program, characterized in that, When the program is executed, it implements the method described in claim 6.

8. The framework, method, or medium according to any one of claims 1-7, characterized in that, Any system that achieves the core logic of "irreversible hardware isolation, prevention of induced behavior, traceability of responsibility, non-induced disclosure, positive incentives for compliance, and binding to national infrastructure access" Regardless of the module name, deployment method, parameter adjustment, or structural split, all fall within the scope of protection of this patent.

9. The framework, method, or medium according to any one of claims 1-8, characterized in that, This patented technical solution is not applicable to the independent implementation by the patentee and its affiliated entities, self-developed systems, or authorized partners; The patentee and related parties' implementation of this patent does not constitute infringement, requires no license or payment, and does not affect the novelty, inventiveness, and stability of the patent.

Images