Device controlled or accessed by artificial intelligence

Abstract

The invention discloses artificial intelligence control equipment. The core of the artificial intelligence control equipment is an emergency stop mechanism. When abnormal operation or potential safety hazards are detected, access or control of artificial intelligence can be quickly interrupted, and user safety and data privacy are guaranteed. The equipment comprises a stop trigger interface (supporting various inputs such as physical / virtual buttons, voice / gesture recognition and the like), a control module (executing basic / emergency stop, such as signal revocation, power failure and the like), a feedback module (providing stop prompt) and a data management module (clearing data / recording state). The method is suitable for various devices such as mobile phones, computers and smart watches, safety risks caused by artificial intelligence abnormal behaviors are effectively prevented, and safety and reliability of the devices are ensured.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence security and control, and particularly to a technical solution for interrupting the artificial intelligence's access to device data and control behavior through an emergency stop mechanism in a device controlled or accessed by artificial intelligence. Background Art

[0002] With the rapid development of artificial intelligence (AI) and large language model (LLM) technologies, AI applications have been widely integrated into various intelligent devices. These devices can perform data analysis, information processing, and decision support through AI, thereby realizing various functions such as automatic ticket booking, intelligent recommendation, and financial operations. However, when AI autonomously accesses device data (such as reading screen content, obtaining user input, or sensor data), it may exhibit abnormal behavior due to algorithm errors, system failures, or external interference, thus posing security risks. For example: a) Some AI applications may malfunction or be maliciously interfered with during automatic ticket booking or automatic transfer processes, resulting in incorrect execution of financial transactions; b) When AI processes data on a device, it may read or leak user privacy data.

[0003] Different from traditional physical devices such as industrial robots, such intelligent devices do not involve physical movement, and their AI behaviors are fully reflected in data access and processing at the software level. Traditional emergency stop buttons and physical power-off methods are difficult to directly apply to such scenarios.

[0004] In addition, as artificial intelligence technology continues to evolve, its capabilities in autonomous decision-making and self-learning are increasing day by day, and potential risks are also rising. Some experts are concerned that if effective control measures are lacking, there may be a risk in the future that AI will pose a threat to human society or even "rule" over humans. To prevent such situations from occurring, it is urgent to introduce security isolation and emergency stop mechanisms in device and system designs to ensure that when AI behavior gets out of control or shows an abnormal self-evolution trend, its control can be interrupted in a timely manner, maintaining humans' ultimate decision-making power and security guarantee over technology.

[0005] The existing technologies mainly focus on the emergency stop of physical movement devices (such as elevators and industrial robots), relying on hardware buttons or relays to cut off the power supply. For AI that operates at the software level, its emergency isolation and stop methods are not yet perfect. Therefore, there is an urgent need for an emergency stop mechanism that can be triggered multimodally (such as physical buttons, virtual buttons, voice, gestures, remote control, and environmental sensors), so as to quickly interrupt the AI's access to or control of the device when detecting abnormal or harmful scenarios, and at the same time provide feedback, data clearing, and status recording to ensure user safety and data privacy. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides a device controlled or accessed by artificial intelligence. The core idea is to set a stop trigger interface and a control module in the device to achieve an emergency interruption of the artificial intelligence's autonomous access to the data in the device (including but not limited to access to the device screen display content, user interaction data, sensor data, application program data, and other relevant information). The specific invention content includes: 1. Stop trigger interface Receive stop instructions sent by users or systems through various input methods (physical buttons, virtual buttons, voice recognition, gesture recognition, remote control, environmental sensors, etc.). In particular, environmental sensors can be used to monitor abnormal situations in scenarios such as financial transactions (for example, when the AI automatically opens online banking for transfer operations, if abnormal transfer risks are detected, a stop instruction is automatically triggered).

[0007] 2. Control module After receiving the stop instruction, the control module takes corresponding measures to interrupt the artificial intelligence's access to or control of the device. The control module includes a basic-level stop unit and / or an emergency-level stop unit: a) The basic-level stop unit aborts the control in a conventional manner by sending a stop instruction to the artificial intelligence or revoking its access permission; b) When detecting an emergency situation, the emergency-level stop unit takes coercive measures (such as closing the local application running the large language model, disconnecting the artificial intelligence processing module connection, shutting down the device processor, or powering off) to immediately abort the control.

[0008] 3. Feedback module and data management module After the stop operation is completed, the feedback module provides a clear prompt to the user (such as "Stop successful" or "Status has been restored"); at the same time, the data management module clears the context data of the artificial intelligence or records the status data at the time of stopping for subsequent security analysis or system recovery.

[0009] To prevent the AI from continuing to access or control the device in special situations, the present invention also provides a configuration option, that is, setting the stop trigger interface or the control module to a continuously active state, so that the device is always in an artificial intelligence isolation state until it is released from isolation by an authorized user. Beneficial effects

[0010] 1. It can quickly interrupt the artificial intelligence's access to or control of the device, preventing potential safety hazards caused by abnormal AI behavior; 2. Meet the needs of different usage scenarios through various triggering methods; 3. Provide feedback and data management at the same time to ensure the transparency of the system state and data security after an emergency stop; 4. It can achieve long-term AI isolation and enhance device security. Description of the Drawings

[0011] Figure 1 : Schematic diagram of the overall structure of the device This figure shows the overall structural framework of a device controlled or accessed by artificial intelligence. All the main modules and interfaces are shown in the figure, and the specific descriptions are as follows: 1. Device main body In the figure, it is represented by the whole "device". The device described in the present invention includes multiple cooperating modules to ensure that after receiving a stop instruction, it can quickly interrupt the access of artificial intelligence to device data or control.

[0012] 2. Stop trigger interface Function and composition: This interface is a key component for the device to receive stop instructions and supports multiple input methods. The detailed input sub-modules are listed in the figure, including: • Physical button: A hardware button used to directly trigger the stop instruction.

[0013] • Virtual button: Located in the graphical user interface for users to operate by touching or clicking.

[0014] • Voice recognition: Triggers the stop operation by capturing voice commands.

[0015] • Gesture recognition: Relies on specific gesture actions as instruction inputs.

[0016] • Remote control interface: Supports instruction inputs from external control devices (such as remote controls or mobile devices).

[0017] • Environmental sensor: Automatically sends a stop signal by monitoring environmental anomalies (such as detecting danger or abnormal operations).

[0018] Working process: Each input method works independently. Once a stop signal is detected, the instruction is immediately transmitted to the central "control module".

[0019] 3. Control module Function: It is the central control unit of the device, responsible for receiving stop instructions and scheduling subsequent modules to perform corresponding operations according to preset logic.

[0020] Configuration option: The note on the right side of the figure marks the "continuous activation state" configuration option, indicating that this module can be configured to remain continuously activated, thus ensuring that the device is always in the AI isolation state under specific circumstances.

[0021] 4. Feedback module Function: After the device successfully interrupts the artificial intelligence access, the feedback module provides clear prompts to the user, such as displaying status information like "Stop successful" or "Device isolated", ensuring that the user can understand the device status in real time.

[0022] 5. Data Management Module Function: Responsible for clearing the context data of the artificial intelligence after the stop operation is executed, or recording the status of the device when it stops, operation logs, and other key data, providing a basis for subsequent security analysis and system recovery.

[0023] 6. Reset / Initial State Module Function: Shown as "Reset / Initial State Module" in the figure, after receiving the instruction from the control module, it restores the device to the preset initial state, ensuring that the device can quickly return to the available state after an emergency interruption.

[0024] 7. Signal Flow Description Each input device (physical button, virtual button, voice recognition, gesture recognition, remote control interface, and environmental sensor) sends a stop signal to the control module.

[0025] After receiving the instruction, the control module triggers the feedback module, data management module, and reset module in sequence to update the device status, record data, and restore the state. Figure 2 : Schematic Diagram of the Detailed Structure of the Control Module Figure 2 : Schematic Diagram of the Detailed Structure of the Control Module This figure focuses on the control module inside the device, and details two core units inside the control module and their specific measures, which are described as follows: 1. Overview of the Control Module Location and Function: The control module, as the core decision-making unit of the present invention, is responsible for receiving the stop signal from the stop trigger interface and deciding what kind of interruption measures to take based on the device status and preset conditions. The figure shows that the control module adopts a hierarchical design inside, divided into a basic-level stop unit and an emergency-level stop unit.

[0026] 2. Basic-Level Stop Unit Function Description: The basic-level stop unit is mainly responsible for interrupting the control behavior of the artificial intelligence in a standard manner when receiving a regular stop instruction.

[0027] Operation Method: By sending a stop signal to the running artificial intelligence application or process, or directly revoking its data access permission, thereby aborting its control of the device.

[0028] Effect: Achieve a quick interruption while ensuring that the device can smoothly transition to a safe state under normal circumstances.

[0029] 3. Emergency Stop Unit Function Description: When the system detects an emergency or abnormal situation (such as an escalation of safety risks, abnormal operations, or consecutive incorrect instructions), the emergency stop unit will be rapidly activated.

[0030] Internal Measures: Multiple specific measures within the emergency stop unit are detailed in the figure, including: • Shut down local applications / processes: Directly shut down the local applications or processes running the artificial intelligence.

[0031] • Disconnect the artificial intelligence processing module: Interrupt the communication link with the core artificial intelligence processing module.

[0032] • Shut down the device processor: Immediately interrupt the operation by shutting down part or all of the functions of the processor.

[0033] • Disconnect the device power supply: Completely stop the device operation by cutting off the power supply in extreme cases.

[0034] Performance Metrics: To ensure a safe response, the emergency measure response time requirement is usually within 50 ms, ensuring the rapid interruption of all AI operations in case of an emergency.

[0035] 4. Internal Connections and Data Flows Although there is no direct data transfer relationship shown in the figure between the two units inside the control module, both independently receive instructions from the central control module and perform corresponding interruption operations externally.

[0036] The emergency stop unit usually serves as a backup solution and is automatically enabled when the basic stop unit fails to respond in a timely manner or when the safety risk is relatively high.

[0037] 5. Redundancy Design and Safety Assurance The note part in the figure emphasizes that a dual signal path or redundancy design is adopted inside the control module to ensure that even if one unit fails, the other unit can still guarantee the normal operation of the safety interruption function, ensuring that the device is always under control. Specific Implementation Modes

[0038] The present invention will be described below in conjunction with specific embodiments, but the protection scope of the present invention is not limited thereby.

[0039] Embodiment 1: Emergency Stop Solution for Smartphones A smartphone is equipped with an internal emergency stop system. When the smartphone runs an AI application (such as providing an automatic ticket booking service using a large language model), the user can trigger a stop instruction in any of the following ways: a) Press the physical emergency stop button preset on the side of the phone; b) Click the virtual stop button displayed on the screen; c) Issue the preset voice command “Stop AI”; d) triggered by a gesture (e.g., a quick swipe on the screen); e) Abnormal operation is detected by the device’s built-in environmental sensors (for example, when AI automatically opens online banking and attempts to transfer money, the system automatically recognizes the abnormality and triggers a stop).

[0040] When any trigger mode receives a stop instruction, the stop trigger interface transmits the signal to the control module.

[0041] The basic-level stop unit in the control module first sends a stop command to the process running the AI ​​application, or revokes its permission to access sensitive data in the mobile phone; If the basic-level measures fail to immediately terminate the operation or there is a special need, the emergency-level stop unit will be activated to achieve immediate interruption by shutting down the AI ​​application, disconnecting the AI ​​processing module, or even disconnecting the power of the mobile phone.

[0042] Subsequently, the feedback module displays prompts such as "AI has stopped" or "System has been isolated" on the mobile phone screen, and the data management module clears the context data of the current AI session and / or records the time, operation log and device status information when it stops. If the user selects long-term isolation mode, the stop trigger interface or control module will remain activated to ensure that the device is in AI isolation until the user manually releases it.

[0043] In this embodiment, data exchange is achieved between the main modules through a dedicated communication bus, and redundant design is adopted to ensure the stability and reliability of the system. The specific details are as follows: 1. Stop triggering interface Composition and connection: Includes physical buttons, virtual buttons on the graphical user interface, voice recognition module, gesture recognition sensor, remote control interface and environmental sensors. Each input device is independently connected to the central control unit.

[0044] Workflow and key parameters: a) When any input device detects a stop signal triggered by the user or the system, it notifies the central control unit through a hardware interrupt or software signal.

[0045] b) Key response time requirements: physical button response time ≤ 100ms, voice and gesture recognition response time ≤ 200ms.

[0046] c) The system has built-in signal filtering algorithm to effectively prevent false triggering and ensure the accuracy of instructions.

[0047] 2. Control module Module Composition: It includes a basic-level stop unit and an emergency-level stop unit.

[0048] Basic-level Stop Unit: a) After receiving a stop instruction, it regularly interrupts the control behavior of the artificial intelligence by sending a stop signal to the running artificial intelligence module or revoking its data access permission.

[0049] b) Ensure that the device can be quickly restored to a safe state under standard operations.

[0050] Emergency-level Stop Unit: When the system detects continuous abnormal data access or security risks (such as unauthorized financial operations), it is immediately activated, and an emergency interruption with a response time ≤ 50 ms is achieved by directly disconnecting the device power supply, shutting down some or all of the processor functions, or interrupting the network connection.

[0051] Connection and Redundancy Design: A dual-signal path design is adopted between the control module and the stop trigger interface, feedback module, and data management module, so that even if one path fails, the other path can ensure the normal response of the system.

[0052] 3. Feedback Module Function Description: After the stop operation is completed, it conveys information such as "Stop Successful" and "System Isolated" to the user in various ways such as screen display, sound prompt, or vibration feedback.

[0053] Adaptability: Automatically adjusts the feedback method according to different device types (such as mobile phones, computers, smart wearable devices) to ensure that users can obtain clear operation status in the shortest time.

[0054] 4. Data Management Module Data Processing: After stopping, immediately clear the context data of the artificial intelligence to prevent the leakage of sensitive information; at the same time, record key data including the time at the stop, operation logs, fault codes, and device status for subsequent security analysis and system recovery.

[0055] Security Assurance: All data transmission and storage adopt encryption algorithms and use secure erasure technology to ensure that data processing complies with current security standards.

[0056] The above modules work together. Through clear hardware and software interface definitions and strict parameter control, the present invention can stably and efficiently interrupt the control of the device by the artificial intelligence in various application scenarios, thereby effectively preventing security risks caused by abnormal behaviors of the artificial intelligence.

[0057] Embodiment 2: Virtual Stop Button Solution for Computer On a computer, the pre-installed AI software runs automatically and can access the user data stored in the computer. A virtual stop button is set on the computer interface. When the user discovers that the AI performs abnormal operations (such as automatically starting sensitive applications in the background, accessing the screen content), a stop instruction is triggered by clicking the virtual button.

[0058] At this time, the stop trigger interface receives the instruction and passes it to the control module.

[0059] The control module uses the basic-level stop unit to send a stop signal to the AI software and immediately revoke its access permission; If the computer detects a more serious risk (such as the AI is performing unauthorized financial operations), the emergency-level stop unit is automatically activated to quickly interrupt the AI operation by disconnecting from the AI processing module or shutting down some processor functions.

[0060] Subsequently, the system pops up a prompt of "AI control has been stopped, please check the system status" on the screen through the feedback module, and at the same time, the data management module records the detailed log and system status at the time of shutdown.

Claims

1. A device controlled or accessed by artificial intelligence, characterized in that, The device includes: (1) A stop trigger interface for receiving a stop instruction; (2) A control module connected to the stop trigger interface, configured to stop the artificial intelligence from accessing or controlling the device when receiving the stop instruction; wherein, the "access" means that the artificial intelligence independently obtains, reads, processes, and / or displays data or information stored or presented in the device.

2. The device according to claim 1, characterized in that, If the device is controlled by artificial intelligence, the control of the device by the artificial intelligence will not cause physical movement of the device itself.

3. The device according to claim 2, wherein, The device is applicable to mobile phones, computers, smart watches, wearable devices, or Internet of Things devices.

4. The device according to claim 1, characterized in that The stop trigger interface includes at least one input method for receiving a stop instruction, and the input method is selected from: (1) A physical button; (2) A virtual button on the graphical user interface; (3) A voice recognition interface; (4) Gesture recognition; (5) A remote control interface; (6) An environmental sensor.

5. The device according to claim 4, characterized in that, The environmental sensor is used to detect situations where it is necessary to stop the artificial intelligence from accessing or controlling the device in a harmful scenario.

6. The device according to claim 1, characterized in that, The control module includes a basic-level stop unit and / or an emergency-level stop unit; (1) The basic-level stop unit is configured to stop the artificial intelligence from controlling the device in a conventional manner when receiving a stop instruction, including sending a stop instruction to the artificial intelligence or revoking its access permission; (2) The emergency-level stop unit is configured to immediately stop the artificial intelligence from controlling the device when detecting an emergency. The emergency-level stop unit includes at least one measure selected from: (a) Closing the running local artificial intelligence application or process; (b) Disconnecting the connection of the artificial intelligence processing module; (c) Shutting down the device processor; (d) Disconnecting the device power supply.

7. The device according to claim 1, characterized in that, The device further includes a feedback module for providing feedback information to the user after the artificial intelligence stops controlling the device. The feedback information includes a prompt message indicating successful stop and / or a prompt message indicating that the device state has been restored.

8. The device according to claim 1, characterized in that, The device further includes a data management module for processing artificial intelligence-related data after the artificial intelligence stops controlling the device. The processing is selected from: (1) Clearing the context data of the artificial intelligence; (2) Recording the state data at the time of stop.

9. The device according to claim 1, characterized in that, After receiving the stop instruction, the device further resumes to the initial state, making the device in an available state.

10. The device according to claim 1, characterized in that, The stop trigger interface or the control module can be configured to be in a continuously active state, so that the device is always in an artificial intelligence isolation state, that is, the artificial intelligence is prohibited from accessing or controlling the device.

Images