Bank anti-fraud method, system and device based on AI and storage medium

Abstract

The embodiment of the invention discloses an AI-based bank anti-fraud method, system and device and a storage medium, and the embodiment of the invention realizes anti-fraud from the perspective of banks by using an AIGC technology, has the advantages of high efficiency, accuracy, profession, complete evidence, safety, reliability and the like, and can effectively prevent fraud and protect customer benefits. Through the synergistic effect of the functions of monitoring call data in real time, analyzing audio emotion, judging fraud behaviors, performing interactive persuasion, storing an evidence chain and the like, a brand new anti-fraud solution is provided for a bank. In future application, optimization and improvement can be continued through continuous machine learning, and a greater contribution is made for fighting fraud crimes. And meanwhile, the security of the data is continuously concerned and enhanced, so that the privacy of the customer and the security of the data are fully protected.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of artificial intelligence technology, and specifically to an AI-based bank anti-fraud method, system, device and storage medium. Background Art

[0002] With the rapid development of information technology, fraud methods are becoming increasingly diverse and intelligent, posing a serious threat to the property safety of bank customers. These fraud methods are constantly evolving, making identification and prevention increasingly difficult. Traditional fraud detection systems are mostly based on single or limited transaction features, such as transaction amount, frequency or time interval. Traditional anti-fraud detection often cannot adapt to complex and diverse fraud patterns, resulting in the inability to effectively identify more clever and hidden fraud methods. In order to effectively prevent fraud and protect the property safety of the people, it is of great practical significance to develop a bank anti-fraud terminal system using AIGC technology. Summary of the invention

[0003] To this end, the embodiments of the present invention provide an AI-based bank anti-fraud method, system, device and storage medium to solve the technical problems that the existing anti-fraud system is unable to professionally interact with suspected fraudsters through dialogue, judge the suspicion of fraud by the other party through questions and answers, analyze the other party's emotional changes through audio, and retain professional audio recording evidence.

[0004] In order to achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0005] According to a first aspect of an embodiment of the present invention, there is provided an AI-based bank anti-fraud method, which is applied to AIGC and includes:

[0006] S1. Comprehensively collect existing fraud patterns to generate training sets, use the training sets to train AIGC for a preset number of times, generate a trained AIGC model, collect bank customers' call data in real time, and use AIGC to analyze the collected audio data in real time to identify voice feature information and emotional change information in the call;

[0007] S2. Use the voice feature information and emotion change information to detect whether the call has a risk of fraud. If there is a risk of fraud, the AI ​​takes over the call to interact with the other party and adjusts the interaction strategy, persuades or stops the other party according to the other party's answer and emotion changes;

[0008] S3. When the call is detected to be over, the entire call process is recorded and associated with the corresponding customer information and fraud identification results to generate a complete chain of evidence and encrypt and store the chain of evidence.

[0009] Furthermore, the call data of bank customers is collected in real time and AIGC is used to analyze the collected audio data in real time to identify the voice feature information and emotional change information during the call, including:

[0010] Collect bank customers' call data through cooperation with operators or bank intranet systems and store them in encrypted form;

[0011] Collect and analyze pitch, speech rate, and volume features from audio data to detect emotional states.

[0012] Further, using the voice feature information and the emotion change information to detect whether the call has a fraud risk includes:

[0013] Using the voice feature information to determine whether there is fraudulent speech;

[0014] If there is fraudulent rhetoric, there is a risk of fraud, and the other party is a suspected fraudster;

[0015] If there is no fraudulent rhetoric, obtaining the other party's emotional state according to the emotional change information;

[0016] Determine whether the other person's emotional state shows signs of nervousness / anxiety;

[0017] If the other party's emotional state shows nervous / anxious emotional characteristics, there is a risk of fraud and the other party is suspected of being a fraudster;

[0018] If the other party's emotional state does not show signs of nervousness / anxiety, there is no risk of fraud.

[0019] Further, using the voice feature information and the emotion change information to detect whether the call has a fraud risk, further comprising:

[0020] Collect fraudulent rhetoric and fraud scenario information and use the fraudulent rhetoric and fraud scenario information to generate fraud patterns and a fraud pattern database;

[0021] By comparing the fraud pattern with the voice feature information, obtaining matching degree numerical information;

[0022] Determine whether the matching degree value information exceeds a preset threshold;

[0023] If the matching degree value information exceeds a preset threshold, an alarm for suspected fraud is triggered;

[0024] If the matching degree value information does not exceed the preset threshold, the fraud pattern database is updated.

[0025] Furthermore, AI takes over the call to interact with the other party and adjusts the interaction strategy, persuades or stops the other party according to their answers and emotional changes, including:

[0026] AI takes over the call and uses professional terms to interact with the other party, obtain the other party's answers and emotional changes, and adjust the interaction strategy based on the other party's answers and emotional changes;

[0027] When it is detected that the other party has a tough attitude, use delaying tactics to buy time while sending information to bank staff so that the bank staff can contact the bank customer.

[0028] Furthermore, using AIGC to perform real-time analysis on the collected audio data also includes:

[0029] Performing voice detection, echo cancellation, sound source localization, and sound enhancement on the audio data to generate an audio signal;

[0030] Utilizing the audio signal to perform speech recognition, language analysis, emotion recognition and event detection to generate corresponding text data;

[0031] The corresponding text data is enhanced by SSML markup language and Prompt and then input into AIGC for analysis.

[0032] Furthermore, the corresponding text data is enhanced by SSML markup language and prompt and then input into AIGC for AI to take over the call and conduct interactive analysis with the other party, including:

[0033] Generate response text and perform speech synthesis to generate corresponding audio information;

[0034] After fine-tuning the sound and color and controlling the emotion of the corresponding audio information, a reply audio signal is synthesized and output as a voice.

[0035] According to a second aspect of an embodiment of the present invention, there is provided an AI-based bank anti-fraud system, the system comprising:

[0036] The data collection module is used to comprehensively collect existing fraud patterns to generate training sets, use the training sets to train AIGC for a preset number of times, generate a trained AIGC model, collect bank customers' call data in real time, and use AIGC to analyze the collected audio data in real time to identify voice feature information and emotional change information in the call;

[0037] An intelligent analysis module is used to detect whether the call has a risk of fraud by using the voice feature information and the emotional change information. If there is a risk of fraud, the AI ​​takes over the call to interact with the other party and adjusts the interaction strategy according to the other party's answer and emotional changes, and conducts persuasion or suppression;

[0038] The evidence preservation module is used to record the entire call process and associate it with the corresponding customer information and fraud identification results when the call is detected to be over, generate a complete evidence chain and encrypt and store the evidence chain.

[0039] According to a third aspect of an embodiment of the present invention, there is provided an AI-based bank anti-fraud device, the device comprising: a processor and a memory;

[0040] The memory is used to store one or more program instructions;

[0041] The processor is used to run one or more program instructions to execute the steps of an AI-based bank anti-fraud method as described in any of the above items.

[0042] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of an AI-based bank anti-fraud method as described in any of the above items are implemented.

[0043] The embodiments of the present invention have the following advantages:

[0044] The embodiment of the present invention uses AIGC technology to achieve anti-fraud from the perspective of the bank, which has the advantages of high efficiency, accuracy, professionalism, complete evidence, safety and reliability, and can effectively prevent fraud and protect customer interests. Through the synergy of functions such as real-time monitoring of call data, analysis of audio emotions, judgment of fraudulent behavior, interactive persuasion and preservation of evidence chains, a new anti-fraud solution is provided for banks. In future applications, the embodiment of the present invention can continue to be optimized and improved through continuous machine learning, making greater contributions to combating fraud crimes. At the same time, data security will also receive continuous attention and strengthening to ensure that customer privacy and data security are fully protected. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0046] The structures, proportions, sizes, etc. illustrated in this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with the technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantial technical significance. Any structural modification, change in proportion or adjustment of size shall still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and purposes that can be achieved by the present invention.

[0047] Figure 1 A logical structure diagram of an AI-based bank anti-fraud system provided in an embodiment of the present invention;

[0048] Figure 2 A flowchart of an AI-based bank anti-fraud method provided by an embodiment of the present invention;

[0049] Figure 3 A functional schematic diagram of an AI-based bank anti-fraud method provided by an embodiment of the present invention;

[0050] Figure 4 A schematic diagram of the architecture of an AI-based bank anti-fraud method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following is a description of the implementation of the present invention by specific embodiments. People familiar with the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Traditional fraud detection systems are mostly based on single or limited transaction features, such as transaction amount, frequency or time interval. Traditional anti-fraud detection often cannot adapt to complex and diverse fraud patterns, resulting in the inability to effectively identify more clever and covert fraud methods. Moreover, current bank anti-fraud is handled by people in person. Bank counters or lobby managers carefully interrogate suspected fraudulent transfer users, then call the police and wait for the police to come to the bank to handle the matter. In this process, the workload of bank staff will increase while increasing the psychological burden. For users, it is also a painful wait.

[0053] In order to effectively prevent fraud and protect the property safety of the people, it is of great practical significance to develop a bank anti-fraud terminal system using AIGC technology.

[0054] In order to solve the technical problem that the existing anti-fraud system is unable to interact with suspected fraudsters through professional dialogue, judge the suspicion of fraud through questions and answers, analyze the emotional changes of the other party through audio, and retain professional audio recording evidence.

[0055] refer to Figure 1 The embodiment of the present invention discloses an AI-based bank anti-fraud system, which includes: a data collection module 1; an intelligent analysis module 2; and an evidence preservation module 3.

[0056] Corresponding to the above-disclosed AI-based bank anti-fraud system, the embodiment of the present invention further discloses an AI-based bank anti-fraud method. The following describes in detail an AI-based bank anti-fraud method disclosed in the embodiment of the present invention in combination with the above-described AI-based bank anti-fraud system.

[0057] The applicable scenarios of the embodiments of the present invention are:

[0058] 1) Bank Customer Service Center

[0059] Deploying this system in the bank's customer service center can monitor customer calls in real time, detect suspected fraud incidents in a timely manner, and intervene. Customer service personnel can further verify and persuade customers based on the system's prompts to protect the customer's property safety;

[0060] 2) Online banking and mobile banking

[0061] By integrating this system into online banking and mobile banking platforms, customers' online transactions can be monitored in real time. When suspected fraudulent transactions are found, the system can promptly issue an alarm to remind customers of the risks and take appropriate measures to intervene;

[0062] 3) Bank branches

[0063] By installing this system in the self-service equipment area of ​​a bank branch, customers' operations on the self-service equipment can be monitored in real time. When suspected fraudulent behavior is detected, the system can automatically sound an alarm to remind bank staff to intervene.

[0064] refer to Figure 2 The present invention discloses an AI-based bank anti-fraud method, which is applied to AIGC and includes:

[0065] S1. Comprehensively collect existing fraud patterns to generate training sets, use the training sets to train AIGC for a preset number of times, generate a trained AIGC model, collect bank customers' call data in real time, and use AIGC to analyze the collected audio data in real time to identify voice feature information and emotional change information in the call.

[0066] The calls are recorded for subsequent audio analysis and evidence preservation. The recorded data is encrypted and stored, and only authorized personnel can access it. The call data of bank customers is monitored in real time, and customer call information is obtained through cooperation with telecom operators or using the bank's internal communication intranet. In the process of data collection, relevant laws, regulations and data security specifications are strictly followed to ensure that customer privacy is fully protected.

[0067] Natural language processing technology is used to extract and analyze keywords in calls. For example, when keywords such as "transfer", "remittance", "winning", "public security, procuratorial and judicial" appear, we will be alert and further analyze the context of the call to determine whether there is suspicion of fraud. The keyword extraction and analysis process uses encryption algorithms to prevent data from being stolen.

[0068] Furthermore, the call data of bank customers is collected in real time and the collected audio data is analyzed in real time using AIGC to identify voice feature information and emotional change information in the call, including: collecting bank customers' call data through cooperation with operators or the bank's communication intranet and encrypting and storing it; collecting and analyzing tone, speaking speed and volume characteristics from audio data to detect emotional status.

[0069] AIGC technology is used to analyze the collected audio data in real time to identify language features, emotional changes, and other information during calls. During the analysis process, advanced encryption algorithms and security measures are used to prevent data from being tampered with or stolen.

[0070] S2. Use the voice feature information and emotional change information to detect whether there is a risk of fraud in the call. If there is a risk of fraud, AI will take over the call to interact with the other party and adjust the interaction strategy according to the other party's answer and emotional changes, and persuade or stop the other party.

[0071] Furthermore, the voice feature information and the emotion change information are used to detect whether there is a fraud risk in the call, including: using the voice feature information to determine whether there is fraudulent talk; if there is fraudulent talk, there is a fraud risk and the other party is a suspected fraudster; if there is no fraudulent talk, the emotional state of the other party is obtained based on the emotion change information.

[0072] Through machine learning algorithms and models, we can determine whether the other party is a fraudster. The specific basis for judgment includes but is not limited to: language patterns, common fraud tactics, emotional fluctuations, etc. The training and updating process of the model is also carried out in a safe environment to ensure the security and reliability of the data.

[0073] Advanced voice emotion recognition technology is used to analyze the emotions of the other party in real time. By analyzing the voice's tone, speaking speed, volume and other characteristics, the other party's emotional state, such as nervousness, anxiety, anger, calmness, etc., is judged. During the emotion analysis process, customer privacy is strictly protected to prevent the abuse of emotion data.

[0074] Determine whether the other party's emotional state shows signs of nervousness / anxiety; if the other party's emotional state shows signs of nervousness / anxiety, there is a risk of fraud and the other party is a suspected fraudster; if the other party's emotional state does not show signs of nervousness / anxiety, there is no risk of fraud.

[0075] Sentiment analysis results can be used as one of the important bases for judging whether the other party is a fraudster. Generally speaking, fraudsters may show emotional characteristics such as nervousness and urgency when talking to customers. At the same time, sentiment analysis can also help adjust interaction strategies and better persuade and stop fraud.

[0076] Furthermore, using the voice feature information and emotion change information to detect whether the call has a fraud risk also includes: collecting fraudulent speech, fraud scenario information and using the fraudulent speech and fraud scenario information to generate fraud patterns and a fraud pattern database; obtaining matching degree numerical information by comparing the fraud pattern with the voice feature information.

[0077] Establish a fraud model database to collect common fraud tactics, fraud scenarios and other information. Through real-time analysis of customer call content, compare it with the fraud model in the database. When a high degree of match is found, an alarm for suspected fraud is triggered. In the process of database management and updating, access rights are strictly controlled to ensure data security.

[0078] Combined with the results of audio emotion analysis and call content analysis, a comprehensive judgment is made as to whether the other party is a fraudster. If there are obvious fraudulent words and abnormal emotions in the call that match the fraud pattern, the other party can be preliminarily judged as a fraudster. In the judgment process, a multi-factor comprehensive evaluation method is used to improve the accuracy and reliability of the judgment.

[0079] Determine whether the matching degree value information exceeds a preset threshold; if the matching degree value information exceeds the preset threshold, trigger an alarm for a suspected fraud event; if the matching degree value information does not exceed the preset threshold, update the fraud pattern database.

[0080] In order to improve the accuracy of judgment, it is necessary to continuously learn and update the fraud model database, and optimize the judgment model through a large number of sample training. At the same time, the embodiment of the present invention can also combine the method of manual review to further confirm suspected fraud incidents. In the process of model training and manual review, data security specifications are strictly followed to ensure the security and confidentiality of data.

[0081] Furthermore, AI takes over the call to interact with the other party and adjusts the interaction strategy, persuades or stops the other party according to the other party's answer and emotional changes, including: AI takes over the call to interact with the other party using professional terms, obtains the other party's answer and emotional changes, and adjusts the interaction strategy according to the other party's answer and emotional changes; when it is detected that the other party has a tough attitude, use delaying tactics to buy time and send information to bank staff at the same time, so that the bank staff can contact the bank customer.

[0082] When AIGC determines that a fraud case is suspected, it automatically takes over the call and interacts with the other party in professional terms. For example, it asks the other party about their identity, purpose, and relationship with the customer in the name of a bank employee. During the interaction, it strictly controls the transmission and storage of information to prevent the leakage of sensitive information.

[0083] According to the other party's answers and emotional changes, the interaction strategy is adjusted to persuade and stop. For example, the legal consequences of fraud can be explained to the other party, and the other party can be reminded not to engage in illegal and criminal activities. The data transmission during the interaction process uses an encrypted channel to ensure information security.

[0084] If the other party is tough, a delaying tactic can be adopted to buy time, so that the bank staff has enough time to contact the customer for further verification and persuasion. During the delaying process, the embodiment of the present invention will continue to monitor the call situation to ensure the property safety of the customer.

[0085] Further, refer to Figure 3 and Figure 4 , using AIGC to perform real-time analysis on the collected audio data, also includes: performing voice detection, echo cancellation, sound source localization, and sound enhancement on the audio data to generate an audio signal; using the audio signal to perform voice recognition, language analysis, emotion recognition, and event detection to generate corresponding text data; and inputting the corresponding text data into AIGC for analysis after being enhanced by SSML markup language and Prompt.

[0086] Furthermore, the corresponding text data is enhanced through SSML markup language and Prompt and input into AIGC for AI to take over the call and conduct interactive analysis with the other party, including: generating response text and performing speech synthesis to generate corresponding audio information; performing voice and emotion fine-tuning and emotion control on the corresponding audio information, and then synthesizing a reply audio signal and performing speech output.

[0087] S3. When the call is detected to be over, the entire call process is recorded and associated with the corresponding customer information and fraud identification results to generate a complete chain of evidence and encrypt and store the chain of evidence.

[0088] The entire call process is fully recorded, including audio and text transcription. During the recording process, a safe and reliable storage method is used to ensure the integrity and traceability of the evidence. The call content is converted into text to facilitate subsequent query and analysis. The call records are associated with relevant customer information, fraud judgment results, etc. to form a complete chain of evidence. The storage and management of the chain of evidence strictly abide by data security specifications to prevent evidence from being tampered with or lost.

[0089] The chain of evidence can serve as an important basis for combating fraud crimes and can be provided to relevant departments for investigation and handling when necessary.

[0090] The embodiments of the present invention have the following advantages:

[0091] 1) Efficiency

[0092] AIGC technology is used to achieve real-time monitoring and analysis, which can detect suspected fraud incidents in a short period of time and take timely intervention measures. Compared with traditional manual monitoring methods, it greatly improves the efficiency of anti-fraud;

[0093] 2) Accuracy

[0094] AIGC technology is used to achieve real-time monitoring and analysis, which can detect suspected fraud incidents in a short period of time and take timely intervention measures. Compared with traditional manual monitoring methods, it greatly improves the efficiency of anti-fraud;

[0095] 3) Professionalism

[0096] Using professional terminology during the interaction process can better communicate with and persuade fraudsters to stop;

[0097] 4) Completeness of evidence

[0098] The complete preservation of the evidence chain during the call provides strong support for combating fraud crimes. The integrity of the evidence chain can ensure that the legal responsibilities of fraudsters can be effectively investigated in subsequent legal procedures;

[0099] 5) Security

[0100] Strict security measures are taken in all aspects such as data collection, analysis, interaction and evidence preservation to ensure customer privacy and data security. The application of technologies such as encrypted storage, encrypted transmission, and access control effectively prevents data from being tampered with, stolen or abused.

[0101] In addition, an embodiment of the present invention also provides an AI-based bank anti-fraud device, which includes: a processor and a memory; the memory is used to store one or more program instructions; the processor is used to run one or more program instructions to execute the steps of an AI-based bank anti-fraud method as described in any of the above items.

[0102] In addition, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of an AI-based bank anti-fraud method as described in any of the above items are implemented.

[0103] In the embodiment of the present invention, the processor may be an integrated circuit chip having the signal processing capability. The processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0104] The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor for execution, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The processor reads the information in the storage medium and completes the steps of the above method in combination with its hardware.

[0105] The storage medium may be a memory, which may be, for example, a volatile memory or a nonvolatile memory, or may include both volatile and nonvolatile memory.

[0106] Among them, the non-volatile memory can be a read-only memory (ROM), a programmable ROM (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory.

[0107] The volatile memory may be a random access memory (RAM) which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDRSDRAM), enhanced synchronous DRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus DRAM (DRRAM).

[0108] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0109] Those skilled in the art will appreciate that in one or more of the above examples, the functions described in the present invention can be implemented using a combination of hardware and software. When software is used, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any media that facilitates the transmission of computer programs from one place to another. Storage media can be any available media that can be accessed by general or special-purpose computers.

[0110] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. An AI-based bank anti-fraud method, characterized in that: The method is applied to AIGC and comprises: S1. Comprehensively collect existing fraud patterns to generate training sets, use the training sets to train AIGC for a preset number of times, generate a trained AIGC model, collect bank customers' call data in real time, and use AIGC to analyze the collected audio data in real time to identify voice feature information and emotional change information in the call; S2. Use the voice feature information and emotion change information to detect whether the call has a risk of fraud. If there is a risk of fraud, the AI ​​takes over the call to interact with the other party and adjusts the interaction strategy, persuades or stops the other party according to the other party's answer and emotion changes; S3. When the call is detected to be over, the entire call process is recorded and associated with the corresponding customer information and fraud identification results to generate a complete chain of evidence and encrypt and store the chain of evidence.

2. The AI-based bank anti-fraud method according to claim 1, characterized in that: Collect bank customers' call data in real time and use AIGC to analyze the collected audio data in real time to identify voice feature information and emotional change information during calls, including: Collect bank customers' call data through cooperation with operators or bank intranet systems and store them in encrypted form; Collect and analyze pitch, speech rate, and volume features from audio data to detect emotional states.

3. The AI-based bank anti-fraud method according to claim 2, characterized in that: Using the voice feature information and the emotion change information to detect whether the call has a fraud risk includes: Using the voice feature information to determine whether there is fraudulent speech; If there is fraudulent rhetoric, there is a risk of fraud, and the other party is a suspected fraudster; If there is no fraudulent rhetoric, obtaining the other party's emotional state according to the emotional change information; Determine whether the other person's emotional state shows signs of nervousness / anxiety; If the other party's emotional state shows nervous / anxious emotional characteristics, there is a risk of fraud and the other party is suspected of being a fraudster; If the other party's emotional state does not show signs of nervousness / anxiety, there is no risk of fraud.

4. The AI-based bank anti-fraud method according to claim 3, characterized in that: Using the voice feature information and the emotion change information to detect whether the call has a fraud risk, further comprising: Collect fraudulent rhetoric and fraud scenario information and use the fraudulent rhetoric and fraud scenario information to generate fraud patterns and a fraud pattern database; By comparing the fraud pattern with the voice feature information, obtaining matching degree numerical information; Determine whether the matching degree value information exceeds a preset threshold; If the matching degree value information exceeds a preset threshold, an alarm for suspected fraud is triggered; If the matching degree value information does not exceed the preset threshold, the fraud pattern database is updated.

5. The AI-based bank anti-fraud method according to claim 4, characterized in that: AI takes over the call to interact with the other party and adjusts the interaction strategy, persuading or stopping the other party based on their answers and emotional changes, including: AI takes over the call and uses professional terms to interact with the other party, obtain the other party's answers and emotional changes, and adjust the interaction strategy based on the other party's answers and emotional changes; When it is detected that the other party has a tough attitude, use delaying tactics to buy time while sending information to bank staff so that the bank staff can contact the bank customer.

6. The AI-based bank anti-fraud method according to claim 5, characterized in that: Using AIGC to analyze the collected audio data in real time also includes: Performing voice detection, echo cancellation, sound source localization, and sound enhancement on the audio data to generate an audio signal; Utilizing the audio signal to perform speech recognition, language analysis, emotion recognition and event detection to generate corresponding text data; The corresponding text data is enhanced by SSML markup language and Prompt and then input into AIGC for analysis.

7. The AI-based bank anti-fraud method according to claim 6, characterized in that: The corresponding text data is enhanced by SSML markup language and prompt and then input into AIGC for AI to take over the call and conduct interactive analysis with the other party, including: Generate response text and perform speech synthesis to generate corresponding audio information; After fine-tuning the sound and color and controlling the emotion of the corresponding audio information, a reply audio signal is synthesized and output as a voice.

8. An AI-based bank anti-fraud system, characterized in that: The system comprises: The data collection module is used to comprehensively collect existing fraud patterns to generate training sets, use the training sets to train AIGC for a preset number of times, generate a trained AIGC model, collect bank customers' call data in real time, and use AIGC to analyze the collected audio data in real time to identify voice feature information and emotional change information in the call; An intelligent analysis module is used to detect whether the call has a risk of fraud by using the voice feature information and the emotional change information. If there is a risk of fraud, the AI ​​takes over the call to interact with the other party and adjusts the interaction strategy according to the other party's answer and emotional changes, and conducts persuasion or suppression; The evidence preservation module is used to record the entire call process and associate it with the corresponding customer information and fraud identification results when the call is detected to be over, generate a complete evidence chain and encrypt and store the evidence chain.

9. An AI-based bank anti-fraud device, characterized in that: The device comprises: a processor and a memory; The memory is used to store one or more program instructions; The processor is used to run one or more program instructions to execute the steps of an AI-based bank anti-fraud method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of an AI-based bank anti-fraud method as described in any one of claims 1 to 7.

Images