Method for assisting in identifying telecommunication fraud through artificial intelligence

Abstract

The invention discloses a method for assisting in identifying telecommunication fraud through artificial intelligence, and relates to the technical field of artificial intelligence risk control, and the method comprises the steps: automatically identifying voice or text content through artificial intelligence, and extracting account and identity information; performing online verification on the account and the identity information, and under the condition that the identity cannot be verified, performing behavior monitoring and evaluating whether the user has a transfer intention or not; when the transfer intention exists, performing communication verification between the artificial intelligence and the cloud server, and if the communication fails or the verification is abnormal, terminating the transaction request; if the communication is normal and the user insists in payment, further judging whether the identity of the payee is credible or not, and if not, automatically transferring to the credible payee account to which the payee belongs; synchronously calling a block chain technology to record transaction information, generating a block chain smart contract, and locking funds to a contract account after user confirmation; and releasing the fund to the collection account after the smart contract condition is fulfilled, otherwise, returning the fund to the user account along the original path.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence risk control technology, and in particular to a method for identifying telecommunications fraud using artificial intelligence assistance. Background Art

[0003] Currently, some smart terminal manufacturers have attempted to deploy anti-fraud model modules, such as "unknown number marking," "suspicious SMS identification," and "pre-payment reminders." However, these measures generally suffer from problems such as "structural fragmentation," "detachment from transaction context," and "lack of data closure," making it difficult to build a controllable interruption chain before the user actually makes a payment. On the other hand, existing technologies are mostly limited to static blacklist rules or keyword matching, and fail to achieve comprehensive judgment based on dynamic dimensions such as the user's actual behavior path, the payee's true identity verification, and transaction context semantic analysis. This leads to a high false alarm rate, delayed prevention and control, and difficulty in covering complex and changing fraud methods. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for identifying telecommunications fraud with the assistance of artificial intelligence to solve the problem of transaction risk control failure caused by non-real-time identity verification, unquantifiable behavioral intentions and lack of controllability of payment paths in telecommunications fraud.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for identifying telecommunications fraud with the assistance of artificial intelligence, which includes: when a user receives an unfamiliar call or text message, the artificial intelligence automatically recognizes the voice or text content and extracts the account and identity information; the extracted account and identity information are verified online, and when the identity cannot be verified, the behavior is monitored and the user is evaluated whether the transfer intention exists; when there is a transfer intention, the artificial intelligence communicates with the cloud server for verification, and terminates the transaction request if the communication fails or the verification is abnormal; if the communication is normal and the user insists on paying, it is further determined whether the identity of the payee is credible, and if it cannot be confirmed, the payment is automatically transferred to the trusted public account to which the payee belongs; during the transfer execution process, blockchain technology is synchronously called to record transaction information. If it is detected that the transaction involves service commitments or rebate inducements, the artificial intelligence generates a blockchain smart contract and locks the funds in the contract account after the user confirms; when the conditions of the smart contract are fulfilled, the funds are released to the receiving account, otherwise the funds are returned to the user's account; the artificial intelligence will feedback the identification, verification, transfer and contract performance information to the user in the form of pictures and texts, and report the key data as evidence to the relevant regulatory platform at the same time.

[0008] As a preferred embodiment of the method for assisting in identifying telecommunications fraud through artificial intelligence according to the present invention, wherein: the artificial intelligence automatically identifies voice or text content and extracts account and identity information therefrom, including the following steps,

[0009] Perform a voice-to-text operation on the call content to obtain a semantic sequence, and extract the account field and the identity field through an entity recognition model;

[0010] For text messages or chat texts, use regular matching and semantic recognition to extract the set of account and identity information fields;

[0011] If image content is received, call OCR for text recognition, and at the same time recognize the QR code target to parse out the collection link or the bound account.

[0012] As a preferred embodiment of the method for assisting in identifying telecommunications fraud through artificial intelligence according to the present invention, wherein: the online verification of the extracted account and identity information is performed, and in the case where the identity cannot be verified, behavior monitoring is performed and it is evaluated whether the user has a transfer intention, including the following steps,

[0013] Submit the extracted account and identity information to the connected official database through an interface;

[0014] Compare whether the fields are consistent. If they are inconsistent, push a "suspicious identity information" prompt to the user, and send the abnormal account to the financial supervision system through the interface;

[0015] The artificial intelligence monitors the user operation behavior in the user's intelligent device and constructs a payment intention vector;

[0016] If the payment intention score exceeds the payment intention determination threshold, it is considered that the user is about to transfer, and proceed to the next verification process; otherwise, no further operation is required.

[0017] As a preferred embodiment of the method for assisting in identifying telecommunications fraud through artificial intelligence according to the present invention, wherein: the communication verification between the artificial intelligence and the cloud server means that the artificial intelligence locally generates a random number and sends it to the cloud server; the cloud server returns the encrypted result, and the artificial intelligence locally decrypts and verifies it.

[0018] As a preferred embodiment of the method for assisting in identifying telecommunications fraud through artificial intelligence according to the present invention, wherein: further determine whether the payee's identity is credible. If it cannot be confirmed, automatically transfer the funds to the affiliated trusted corporate account, including the following steps,

[0019] Judge whether the receiving account belongs to an authenticated platform, institution or corporate white list account;

[0020] If the receiving account has no authoritative authentication, transfer the funds to the trusted corporate account associated with it;

[0021] The receiving account is controlled by the bank or the platform. After that, the user can apply to confirm the information before performing payment or refund.

[0022] As a preferred solution of the method for assisting in identifying telecommunications fraud through artificial intelligence according to the present invention, wherein: the invocation of blockchain technology to record transaction information means that when the user makes a payment, a structured data block including the payer's account number, payee's account number, payment amount, timestamp, identity verification result, and behavior intention score is generated for the transaction content and written into the blockchain ledger to form an immutable record.

[0023] As a preferred solution of the method for assisting in identifying telecommunications fraud through artificial intelligence according to the present invention, wherein: when the smart contract conditions are fulfilled, the funds are released to the receiving account; otherwise, the funds are returned to the user's account along the original path, including the following steps.

[0024] Construct a contract structure and lock the funds in the contract account, waiting for the conditions to be triggered.

[0025] If the contract conditions are met, the funds are released; if the conditions are not completed within the time limit, the funds are returned to the user's account along the original path.

[0026] As a preferred solution of the method for assisting in identifying telecommunications fraud through artificial intelligence according to the present invention, wherein: the synchronization and reporting of key data as evidence to relevant regulatory platforms means that artificial intelligence packages the voice transcription content, field extraction results, identity verification logs, user behavior scores, transaction chain information, and contract execution results during the identification process to form structured risk evidence data, and uploads it to the financial regulatory agency and the anti-fraud public security platform through an interface.

[0027] In a second aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and wherein: when the computer program is executed by the processor, any step of the method for assisting in identifying telecommunications fraud through artificial intelligence as described in the first aspect of the present invention is implemented.

[0028] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and wherein: when the computer program is executed by the processor, any step of the method for assisting in identifying telecommunications fraud through artificial intelligence as described in the first aspect of the present invention is implemented.

[0029] The beneficial effects of the present invention are: through online cross-verification of account and identity fields and combined with user behavior intention recognition, dual recognition of identity forgery and induced payment fraud incidents is achieved, thereby improving the intervention depth and judgment accuracy in the non-payment stage; through the random number generation and encryption response mechanism used between artificial intelligence and cloud servers, the authenticity of the user's local operation behavior is confirmed, avoiding the user terminal from being illegally controlled to make payments without perception, and effectively combating technical fraud methods that control equipment through remote control tools. It is a technical loophole that traditional payment interception methods cannot cover, and has significant identification boundary expansion significance; by judging the credibility of the receiving account and setting up a transit mechanism, the funds of suspicious collection paths are isolated and controlled, thereby ensuring that funds do not directly flow into illegal accounts in high-risk scenarios, and gaining time for users to remedy the situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 This is an overall flow chart of the method for identifying telecommunications fraud through artificial intelligence assistance in Example 1.

[0032] Figure 2 This is a flow chart of identity information extraction and verification in Example 1.

[0033] Figure 3 This is a flowchart of the smart contract processing in Example 1.

[0034] Figure 4 This is a flowchart of evidence reporting and user feedback in Example 1. DETAILED DESCRIPTION

[0035] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0036] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0037] Second, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other with other embodiments.

[0038] Embodiment 1, referring to Figures 1 to 4 , this embodiment provides a method for identifying telecom fraud assisted by artificial intelligence, including the following steps:

[0039] S1. When the user receives an unfamiliar call or short message, the artificial intelligence automatically identifies the voice or text content and extracts the account number and identity information.

[0040] When the user device (such as a smart phone) receives an unfamiliar call or short message, the artificial intelligence module is automatically activated. The artificial intelligence extracts the account information (such as bank card number, mobile phone number) and identity information (such as name, work unit) from the conversation content or text through the speech recognition (Speech-to-Text, STT) and natural language processing (NLP) models.

[0041] Specifically, for a voice call: the artificial intelligence calls the speech recognition (ASR) module to transcribe the call content into a sequence of semantic fragments T = {s1, s2,..., s n}, each fragment representing a discourse interaction between the user and the caller;

[0042] The natural language processing (NLP) module is called for the transcribed text to perform named entity recognition (NER) and keyword matching to extract the following information fields:

[0043] Account information: bank card number, collection QR code number, and payment platform account;

[0044] Identity information: person's name, affiliated unit, professional identity, organization name, etc.;

[0045] Commitment information: semantic fragments containing sentence patterns such as "task cashback", "immediate money back after completing the operation", "investment will surely get".

[0046] For a text message: the text message content is clause-separated, de-noised, and structurally analyzed; the pattern matching algorithm is called, and regular expressions and semantic templates are used to identify the three types of information: account information, identity information, and commitment information; for short link or URL fields (such as "pay.xxx.com / abc123"), it is further determined whether it jumps to a payment or real-name authentication platform.

[0047] If it is image information (such as a screenshot of a payment QR code, a chat screenshot): The artificial intelligence calls the OCR module to extract the text in the image; if the recognized content contains a QR code, the QR code information is decoded to extract the potential payment link or the payee's account number; at the same time, the text in the image is scanned to identify whether there is account information, identity information, and commitment information.

[0048] Tag all the extracted fields and construct a structured tuple:

[0049]

[0050] Among them, f i represents the specific field text extracted; type i represents the field type, including Account, Identity, Commitment; ω i represents the confidence score assigned by the AI model according to the context, source, and expression way, and m represents the total number of field entries extracted from the structured field set.

[0051] Conduct a preliminary screening of the structured fields. If an Account or Commitment field is recognized, or the field comes from an unfamiliar call and does not match the user's address book, then conduct the next verification and anomaly identification of the account and identity information.

[0052] S2. Conduct an online verification of the extracted account and identity information, and in the case where the identity cannot be verified, conduct behavior monitoring and evaluate whether the user has the intention to transfer funds.

[0053] Specifically, extract all Account type fields A = {a1, a2,..., a m} and Identity type fields I = {i1, i2,..., i n} from the structured field set F, and construct verification pairs:

[0054] The artificial intelligence sends each group of verification pairs to the public security organ's population database, the bank's real-name account system, and the communication real-name system through the API interface for verification, and the returned results are as follows:

[0055]

[0056] Among them, R jk is whether the field pair is verified as a binding relationship by the database, and the value is 0 or 1; 1 means there is a binding, 0 means there is no binding; i k is the kth identity field, a j the jth account field.

[0057] For each verification field pair Qjk , combined with the field source confidence ω jk (depending on the information source, identification method, and keyword confidence level) calculate the matching score M jk :

[0058] M jk = ω jk ·R jk ;

[0059] Calculate the average matching score G for all matches:

[0060]

[0061] where |Q| is the total number of all field pairs.

[0062] If G < θ (e.g., θ = 0.6), then determine that the communication event is "abnormal identity information", and perform the following operations in sequence: display a pop-up window to the user, clearly prompt: "The identity of the caller cannot be verified. It is recommended to abort the conversation and avoid transferring money"; pack all the involved account information, field pairs with matching failures, recognition texts, etc. into structured risk control data and push it to the bank risk control system and the public security anti-fraud platform in real time; at the same time, encrypt and archive the original communication data (voice transcription text, SMS, image), and generate a digital fingerprint summary for subsequent evidentiary purposes.

[0063] If G ≥ θ, then do not issue a warning for the time being, and the system continues with the payment intention monitoring and judgment process.

[0064] Furthermore, in the case of failed identity verification or significant uncertainty (i.e.), it is necessary to further determine whether the user is performing a suspicious payment operation to assist in the decision of whether to continue tracking the transaction behavior.

[0065] Real-time monitor the payment-related operation behaviors on the user's device, specifically including:

[0066] Whether the barcode scanning interface is called or the camera is started (for scanning the payment barcode);

[0067] Whether the payment code (such as WeChat, Alipay QR code) is generated, displayed, or entered;

[0068] Whether the front interface of typical payment applications such as WeChat Pay, Alipay, mobile banking, and online banking APP is opened or stayed;

[0069] Whether key transaction information such as the receiving account, payment amount, verification code, or payment password is manually entered in the above applications;

[0070] Whether an operation action with a payment confirmation function such as fingerprint recognition, face recognition, or gesture password input is performed on the payment terminal.

[0071] Construct a payment behavior feature vector based on the above behaviors:

[0072] where x l is a boolean variable indicating whether the l-th payment behavior occurs. If it occurs, it is 1; otherwise, it is 0. x1 indicates whether the scanning code behavior occurs; x2 indicates whether the payment code generation or input behavior occurs; x3 indicates whether the payment application is started and in an active state; x4 indicates whether the user enters sensitive fields such as account / amount / password, etc.; x5 indicates whether the biometric recognition / confirmation behavior is executed.

[0073] Each value is either 1 or 0, indicating whether the behavior occurs within the observation window.

[0074] Set a group of weights w related to the importance of behaviors l , and calculate the payment intention score S intent :

[0075]

[0076] where w l represents the weight value of the l-th payment behavior, satisfying ∑w l = 1. For example, it can be set that w1 = 0.15, w2 = 0.1, w3 = 0.25, w4 = 0.3, w5 = 0.2;

[0077] Set a payment intention determination threshold (such as 0.5). If the payment intention score exceeds the payment intention determination threshold, it is considered that the user is about to transfer, and proceed to the next verification process; otherwise, no further operation is required.

[0078] S3. When there is a transfer intention, the artificial intelligence communicates with the cloud server for verification. If the communication fails or the verification is abnormal, the transaction request is terminated.

[0079] Specifically, the artificial intelligence locally generates a random number r u , and sends it to the cloud server; the cloud server returns the encrypted result, and the artificial intelligence decrypts and verifies it locally.

[0080] Furthermore, a one-time random number is generated locally by the artificial intelligence as a verification seed, which is used to verify the device uniqueness and local operation consistency;

[0081] The artificial intelligence locally sends r u , the user unique identifier Token, and the current timestamp t to the cloud server together. The cloud server calculates the signature function r s :

[0082] r s = f(r u , Token, t);

[0083] Among them, f() is a preset encryption hash function that supports device verification;

[0084] The cloud server returns the calculation result r s to the local, and uses the same algorithm to encrypt the local r u and compare it with the result returned by the cloud server. If it satisfies: f(r u , Token, t) = r s , it is determined that the communication path is secure; otherwise, it is regarded as communication anomaly or hijacking.

[0085] If the verification fails, immediately abort the payment operation and prompt "Communication verification failed, do not transfer money"; at the same time, record the event in the log system and report it to the cloud risk module;

[0086] If the verification is successful, it is confirmed that the user is currently in a trusted device environment.

[0087] It should also be noted that when an electronic transaction transfer is made to the user's account. When a transfer operation occurs, the AI APP and the cloud server perform random number encryption and decryption operations. If the communication is normal and the encryption and decryption results are correct, then the cloud server will allow this transfer transaction. However, if the AI APP and the cloud server cannot communicate normally, or cannot perform encryption and decryption operations normally, then, before the user's account makes an electronic transfer payment, notify the cloud server, and the cloud server will send a request to terminate the transaction to financial departments such as banks. That is, when the user transfers money, financial institutions such as banks need to receive the transfer information sent by the user and also receive the verification information from the cloud server to perform the electronic transfer payment operation. Usually, for intelligent devices installed with the AI APP, sufficient time is left during the transfer process for the cloud server to report information.

[0088] S4. If the communication is normal and the user insists on making a payment, further determine whether the payee's identity is trustworthy. If it cannot be confirmed, automatically transfer the money to the trusted corporate account to which it belongs.

[0089] Judge whether the receiving account belongs to an authenticated platform, institution, or corporate white list account;

[0090] Specifically, verify whether the receiving account has completed real-name authentication registration and whether it is bound to a legal individual or enterprise; query whether the receiving account comes from the account list of a registered platform or government unit; judge whether the account number is the corporate account of a registered enterprise and whether it exists in the industrial and commercial registration data.

[0091] If the receiving account meets any one of the criteria, it is regarded as a trusted account and the payment can continue; otherwise, enter the payee attribution tracking mechanism and transfer the funds to the trusted corporate account associated with it.

[0092] Furthermore, if the receiving account fails the credibility verification, the system will call the enterprise entity interface of the public security organ and the National Enterprise Credit Information Publicity System to conduct a reverse check on whether the account is associated with a specific organization or unit.

[0093] If it can be determined that the account belongs to a specific registered enterprise, an organizational attribution structure will be constructed:

[0094] E = {entityname, registrationcode, legalaccount};

[0095] And transfer the originally planned payment amount to its legal corporate account: A * = E.legal_account.

[0096] If the legal entity cannot be traced, the transaction will be aborted, and the user will be prompted with "The identity of the payee cannot be confirmed. It is recommended to contact the official platform for handling."

[0097] The receiving account will be controlled by the bank or the platform. The user can apply to confirm the information later and then execute the payment or refund. For all transactions that have completed the trusted account payment or white list transfer operation, the following information will be recorded in the log and the on-chain evidence structure: receiving account (original and final); transfer amount and timestamp; credibility assessment path (real-name authentication result, attribution entity information, transferred account type); transaction behavior and source reference (whether through call instructions, SMS jumps, APP inductions, etc.).

[0098] S5. During the transfer execution process, blockchain technology is synchronously called to record transaction information. If it is detected that the transaction involves service commitments or rebate inductions, artificial intelligence generates a blockchain smart contract and locks the funds into the contract account after the user confirms.

[0099] Calling blockchain technology to record transaction information means generating a structured data block for the transaction content when the user makes a payment and writing it into the blockchain ledger to form an immutable record.

[0100] Specifically, construct a contract structure and build a transaction data structure block B, which contains the following fields:

[0101] B = {payer, payee, amount, t, Mat, S intent , Ver, Ent, Rou};

[0102] Among them, payer is the payment account; payee is the receiving account (if it is a transfer, it is the white list or corporate account); amount is the payment amount; t is the timestamp; Mat is the average matching score of identity verification; S intentIt is for payment intention scoring; Ver is the communication verification status with the cloud server (success / failure); Ent is the structural information of the payee's affiliated entity; Rou is the type of fund flow path (e.g., original receipt, trusted transfer, escrow contract).

[0103] Write the transaction data structure block B into the on-premises consortium blockchain ledger, and complete block packaging and chain-up through a distributed consensus algorithm to ensure that the transaction data cannot be changed once written.

[0104] At the same time, generate a hash digest for this data block: H B = SHA256(B);

[0105] Among them, H B is the unique hash fingerprint of the corresponding transaction block, which is used to compare the payment chain during subsequent contract performance verification, retrieve specific records during user or police investigation, and identify the original payment evidence by the court or judicial department.

[0106] After the transaction information is registered on the blockchain, if it is identified that the user's current payment behavior involves inducement commitment statements such as platform rebates, task cashbacks, shopping refunds, and prepaid benefits, it will enter the contract locking control process.

[0107] Specifically, perform semantic analysis on the speech recognition transcription result, SMS text, and payment page interaction content. If it contains inducement commitment keywords (such as "cashback for completing the operation", "money back after task completion", "instant reward", "certain rebate"), then generate the following blockchain contract structure:

[0108] C = {payer, payee, locked, cond, timeout, releaseRule, hashRef};

[0109] Among them, lockedAmount is the amount to be paid but to be frozen; condition is the payment unlocking condition (such as delivery, receipt, verification confirmation); timeout is the longest waiting time window (such as within 30 minutes or 24 hours); releaseRule is the contract release rule (release the loan if the condition is met, otherwise refund); hashRef is the reference to the transaction data hash H B as the source transaction binding.

[0110] After the contract template is generated, pop up a prompt on the user payment terminal: "It is detected that the current payment involves a commitment transaction, and the funds will be temporarily locked. Please confirm the release by the platform after the commitment is completed."

[0111] After the user confirms, the funds will be frozen in the contract escrow account, and the status is marked as "being locked".

[0112] Continuously monitor the set performance conditions. If it is detected that the service has been completed (such as task submission, delivery notice, QR code verification, etc.), or a receipt is received, the following judgment shall be executed:

[0113]

[0114] It should be noted that the logic of automatically releasing or returning funds is executed through the intelligent contract judgment path, and the entire contract process and status changes are also synchronously recorded in the blockchain ledger for subsequent behavior verification, compliance traceability, or judicial evidence collection.

[0115] S6. When the intelligent contract conditions are fulfilled, release the funds to the receiving account; otherwise, return the funds to the user's account along the original path.

[0116] Lock the funds in the contract account and wait for the conditions to be triggered; if the contract conditions are met, release the funds; if the conditions are not completed within the time limit, return the funds to the user's account along the original path.

[0117] Specifically, continuously monitor whether the set performance conditions in the contract are met. If condition = releaseRule is met within the specified time timeout, the payment release will be automatically triggered; if the conditions are not met within the time limit or a performance failure is detected, it will be returned to the payer along the original path.

[0118] S7. The artificial intelligence will feedback the identification, verification, transfer, and contract performance information to the user in the form of pictures and texts, and synchronously report the key data as evidence to the relevant regulatory platforms.

[0119] According to the payment behavior path, provide multi-channel identification result feedback on the user interface, including: pop-up a prompt on the payment page "Payment risk has been evaluated, and anti-fraud control has been executed for this fund"; if the transaction is locked or transferred to a corporate account, display the specific transfer path and reason explanation; if the final payment is returned, prompt "The payment recipient cannot be verified, and the payment has been withdrawn for you"; synchronously convey the transaction verification and intelligent contract status through SMS, APP push, or device system notification bar.

[0120] The artificial intelligence will package the voice transcription content, field extraction results, identity verification logs, user behavior scores, transaction chain information, and contract performance results during the identification process to form structured risk evidence data, and upload the structured risk evidence data through the interface to: bank risk control platform interface (such as intelligent anti-fraud gateway), public security anti-fraud center data channel, and blockchain traceability node repeater (for subsequent compliance inspection, evidence collection, or rollback).

[0121] This embodiment also provides a computer device, which is applicable to the situation of the method for assisting in identifying telecom fraud through artificial intelligence, and includes: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for assisting in identifying telecom fraud through artificial intelligence as proposed in the above embodiment.

[0122] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the outer shell of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0123] This embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the method for assisting in identifying telecom fraud through artificial intelligence as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read Only Memory (EPROM for short), Programmable Read-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0124] In summary, the present invention realizes the structured capture of potential fraud elements in unfamiliar communications by extracting account and identity information from voices or texts through artificial intelligence, thereby enhancing the ability to perceive initial fraud attempts and effectively solving the problem that the existing system "cannot actively perceive high-risk content" in the early stage of fraud recognition. By online cross-verifying the account and identity fields and combining with the recognition of user behavior intentions, the present invention realizes the dual recognition of fraud events such as identity forgery and induced payment fraud, thereby enhancing the intervention depth and judgment accuracy in the unpaid stage and effectively preventing fraudsters from bypassing the account real-name mechanism to implement fraud. By using the random number generation and encrypted response mechanism between artificial intelligence and the cloud server, the present invention realizes the authenticity confirmation of the user's local operation behavior, avoids the illegal control of the user terminal to implement payment without perception, and effectively combats the technical fraud means of controlling the device through remote control tools, which is a technical loophole that cannot be covered by traditional payment interception means and has a significant meaning of expanding the recognition boundary. By judging whether the receiving account is a real-name authenticated user or a white-list corporate account, if the identity cannot be verified, the funds are automatically transferred to the trusted corporate account of its affiliated institution as a transfer and escrow path, realizing the fund isolation control of the suspicious receiving path, thereby ensuring that the funds do not directly flow into illegal accounts in high-risk scenarios and buying time for users to take remedial measures.

[0125] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for identifying telecommunications fraud assisted by artificial intelligence, characterized in that: including When the user receives an unfamiliar call or text message, the artificial intelligence automatically identifies the voice or text content and extracts the account and identity information; Online verification is performed on the extracted account and identity information, and in the case where the identity cannot be verified, behavior monitoring is performed and it is evaluated whether the user has the intention to transfer funds; When there is an intention to transfer funds, the artificial intelligence communicates and verifies with the cloud server. If the communication fails or the verification is abnormal, the transaction request is terminated; If the communication is normal and the user insists on making a payment, it is further determined whether the payee's identity is trustworthy. If it cannot be confirmed, the funds are automatically transferred to the trusted corporate account to which it belongs; During the transfer execution process, the blockchain technology is synchronously called to record the transaction information. If it is detected that the transaction involves service commitments or rebate inducements, the artificial intelligence generates a blockchain smart contract and locks the funds into the contract account after the user confirms; When the conditions of the smart contract are fulfilled, the funds are released to the payee's account, otherwise the funds are returned to the user's account along the original route; The artificial intelligence feeds back the identification, verification, transfer, and contract execution information to the user in the form of pictures and texts, and synchronously reports the key data as evidence to the relevant regulatory platforms.

2. The method for identifying telecom fraud assisted by artificial intelligence according to claim 1, characterized in that: The artificial intelligence automatically identifies the voice or text content and extracts the account and identity information therein, including the following steps Perform a voice-to-text operation on the call content to obtain a semantic sequence, and extract the account field and identity field through an entity recognition model; For text messages or chat texts, use regular matching and semantic recognition to extract the set of account and identity information fields; If image content is received, call OCR for text recognition, and at the same time identify the QR code target and parse out the collection link or bound account.

3. The method for identifying telecommunications fraud assisted by artificial intelligence according to claim 2, wherein: The online verification of the extracted account and identity information, and in the case where the identity cannot be verified, behavior monitoring is performed and it is evaluated whether the user has the intention to transfer funds, including the following steps Submit the extracted account and identity information to the connected official database through the interface; Compare whether the fields are consistent. If they are inconsistent, push a "suspicious identity information" prompt to the user, and send the abnormal account to the financial supervision system through the interface; The artificial intelligence monitors the user operation behavior in the user's smart device and constructs a payment intention vector; If the payment intention score exceeds the payment intention determination threshold, it is considered that the user is about to transfer funds and proceed to the next verification process; otherwise, no further operation is required.

4. The method for identifying telecommunications fraud assisted by artificial intelligence according to claim 3, wherein: The communication verification between the artificial intelligence and the cloud server means that the artificial intelligence locally generates a random number and sends it to the cloud server; The cloud server returns the encrypted result, and the artificial intelligence decrypts and verifies it locally.

5. The method for identifying telecom fraud through artificial intelligence assistance according to claim 4, characterized in that: The further determination of whether the payee's identity is trustworthy. If it cannot be confirmed, the funds are automatically transferred to the trusted corporate account to which it belongs, including the following steps Judge whether the collection account belongs to an authenticated platform, institution, or corporate white list account; If the collection account has no authoritative authentication, transfer the funds to the trusted corporate account associated with it; The collection account is controlled by the bank or platform, and the user can apply to confirm the information later and then execute the payment or refund.

6. The method for identifying telecommunications fraud assisted by artificial intelligence according to claim 5, wherein: The call to the blockchain technology to record transaction information means that when the user makes a payment, a structured data block including the payer's account number, payee's account number, payment amount, timestamp, identity verification result, and behavior intention score is generated for the transaction content and written into the blockchain ledger to form an immutable record.

7. The method for identifying telecommunications fraud assisted by artificial intelligence according to claim 6, characterized in that: When the smart contract conditions are fulfilled, the funds are released to the payee's account; otherwise, the funds are returned to the user's account along the original path, including the following steps: Construct the contract structure and lock the funds in the contract account, waiting for the conditions to be triggered; If the contract conditions are met, the funds are released; if the conditions are not completed within the timeout period, the funds are returned to the user's account along the original path.

8. The method for identifying telecom fraud assisted by artificial intelligence according to claim 7, characterized in that: The synchronization and reporting of the key data as evidence to the relevant regulatory platforms means that the artificial intelligence packs the voice transcription content, field extraction results, identity verification logs, user behavior scores, transaction chain information, and contract performance results during the recognition process into structured risk evidence data and uploads it to the financial regulatory agency and the anti-fraud public security platform through an interface.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for assisting in identifying telecommunications fraud through artificial intelligence according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for assisting in identifying telecommunications fraud through artificial intelligence according to any one of claims 1 to 8.

Images