AI Fake Customer Service Call Scam — Impersonating platform customer service using AI voice outbound calls to induce money transfers

Who Gets Targeted

Primary victims are middle-aged and elderly consumers who frequently use e-commerce platforms, courier services, and online travel bookings, as well as ordinary online shoppers who naturally trust official customer service identities. Fraud rings utilize leaked order data purchased from the dark web to precisely recite the victim's recent shopping records, tracking numbers, shipping addresses, and other details during the call, causing victims to let down their guard within seconds. Psychological vulnerabilities focus on three aspects: a conditioned reflex of trust toward official customer service titles, panicked responses to keywords concerning account security or financial losses, and a lack of secondary verification awareness driven by a sense of urgency. Living-alone seniors and consumers who have just undergone large-scale consumption are particularly vulnerable.

骗局怎么运作

• Data Acquisition Phase: Fraud rings purchase personal user data leaked from channels such as e-commerce platforms, courier companies, and hotel reservation systems via the dark web or instant messaging groups. This includes precise information such as names, mobile phone numbers, recent order details, and shipping addresses. Some rings also use web scrapers to automatically collect public information to supplement victim profiles. This data enables fraudsters to read out real consumption records line by line during calls, quickly establishing a trust endorsement.
• AI Outbound System Deployment: Fraudsters use AI voice outbound platforms circulating in the grey market, configuring standardized script templates simulating major well-known platforms' customer service. The system supports automatic batch dialing, capable of making thousands or even 1,500 calls a day, far exceeding human customer service efficiency. Voice models are trained on specific scenarios to simulate natural dialogue rhythms, including tone pauses and colloquial expressions, making it difficult for victims to distinguish authenticity immediately.
• Script Induction and Trust Building: Once connected, the AI voice claims to be platform customer service, using leaked data on hand to itemize the victim's real order information, and then claims that the order has anomalies, products need to be recalled and refunded, or accounts involve violations requiring security verification, deliberately creating a sense of urgency. When victims raise questions, the AI system can match responses in real-time based on preset script trees, with some systems already possessing large-model-driven multi-turn dialogue capabilities that bring conversational fluency close to that of a real person.
• Transfer to Human Deep Manipulation: When the AI system determines that the victim has initially taken the bait and shown a willingness to cooperate, it automatically transfers them to a human fraudster, usually under titles such as senior specialist or claims supervisor. Once the human fraudster takes over, they induce the victim to download designated meeting or remote assistance apps to enable screen-sharing functions, or request transfers to so-called secure accounts for verification under the guise of fund security protection. This stage combines technical disguise with human social engineering to actually complete the theft of funds.
• Fund Transfer and Laundering: Human fraudsters peek at the victim's mobile banking operation interface and SMS verification codes in real-time through screen sharing, quickly completing transfers or loan extraction operations. In some cases, fraudsters also arrange personnel to collect cash in person to evade bank transfer tracking. Once funds arrive, they are immediately dispersed and transferred through multi-tier accounts or exchanged for cryptocurrencies to launder the proceeds, spanning multiple payment channels and making subsequent tracking and freezing extremely difficult.
• Number Spoofing and Anti-Tracking: Throughout the entire process, fraud rings use caller ID spoofing software to disguise incoming caller IDs as official numbers starting with 400 or numbers highly similar to real platform customer service hotlines. Outbound numbers are changed frequently, being used and discarded after being purchased in bulk from virtual network operator SIM cards. Some systems have the function of detecting whether the target phone has installed the Anti-Fraud Center app; once detection identifies installation, it automatically skips that number to choose targets that are easier to compromise.

红旗信号（看到这些快跑）

• 🚩 Incoming calls claim to be platform customer service or claims specialists, but the number is an unfamiliar mobile number or virtual number block, rather than the customer service hotline officially published by the platform
• 🚩 The other party can accurately state your name, order number, shipping address, and other real information, but then asks you to confirm your bank card number or provide an SMS verification code
• 🚩 The call requests the download of specific meeting or remote assistance apps to enable screen-sharing or video verification — actions that do not exist in any regular customer service workflow
• 🚩 Creating a sense of urgency using reasons such as order abnormalities, unqualified product quality testing requiring refunds, or accounts suspected of violations requiring unfreezing, urging you to operate immediately and refusing to let you hang up
• 🚩 The AI voice exhibits unnatural repetitions, irrelevant answers, or abnormal pauses when handling complex follow-up questions, noticeably differing from the impromptu response rhythm of real customer service
• 🚩 The other party refuses to let you hang up and call back official customer service independently for verification, or claims that the system is processing and the call cannot be interrupted
• 🚩 Requesting funds to be transferred into so-called secure accounts or verification accounts, or requesting cash withdrawals to be handed over to an on-site specialist for collection

真实案例

• On April 28, 2026, Xinhua News Agency's Xinhua Investigation reported 'Beware of AI Customer Service Call Scams': Aunt Wang in Jinchang City, Gansu Province, received a call from someone claiming to be official customer service of a short-video platform, stating her account had subscribed to a live-streaming membership automatically deducting 800 yuan per month. Aunt Wang followed instructions to download an app, after which over 100,000 yuan in her bank card was transferred away. The Jinchuan Branch of the Jinchang Municipal Public Security Bureau established a task force to arrest 8 criminal suspects, with over 20 victims in this case. (Source: [https://www.news.cn/legal/20260428/ccc23e82b9574c7ca1a641d0d10bbf26/c.html](https://www.news.cn/legal/20260428/ccc23e82b9574c7ca1a641d0d10bbf26/c.html))
• On July 1, 2026, Fuzhou News Network reported that Minhou Public Security cracked down on an AI voiceprint scam: Lawbreakers impersonated the victim's 'grandson', crying on the phone about being investigated by the 'police' for suspected illegal activities, claiming it would take 60,000 yuan to smooth things over, and begging his grandmother to first take out 20,000 yuan as a 'security deposit' for emergency rescue while instructing her to keep quiet. The victim was intercepted and advised in time while rushing to the bank with 20,000 yuan in cash. (Source: [https://news.fznews.com.cn/fzyw/20260701/vUn4p587FK.shtml](https://news.fznews.com.cn/fzyw/20260701/vUn4p587FK.shtml))
• In February 2026, media reported an AI voice-mimicking fraud case. The defendant W used AI voice-mimicking technology to impersonate a victim's family member's voice to commit telephone fraud, and arranged personnel to collect cash illicit gains in person to evade bank transfer tracking channels. The court ruled that their behavior constituted the crime of fraud and sentenced them to criminal penalties. This case is one of the earlier disclosed composite fraud cases involving AI voice simulation combined with on-site cash collection in public reports.
• On August 5, 2026, CNR.cn, citing CCTV News client reports, noted that over 90% of fraud cases are implemented relying on network tools such as apps and fraudulent websites; Du Wei, Deputy Director of the Anti-Fraud Center at the China Academy of Information and Communications Technology (CAICT), introduced that nearly 200 million high-risk fraudulent phone cards and over 400 million internet accounts have been cumulatively investigated and disposed of. In one case, over 20,000 yuan already paid by Mr. Du was refunded by the platform, and the remaining 100,000+ yuan in his account was protected and payment-stopped. (Source: [https://www.cnr.cn/mspd/zhsh/20260805/t20260805_527748019.shtml](https://www.cnr.cn/mspd/zhsh/20260805/t20260805_527748019.shtml))

Official Stance

• On April 28, 2026, Xinhuanet published a special investigative report titled 'Beware of AI Customer Service Call Scams', reminding consumers to remain vigilant against incoming calls claiming to be platform customer service and not to easily trust rhetoric about order refunds or account abnormalities
• On August 5, 2026, CNR.cn jointly with anti-fraud departments reported that over 90% of scams are hidden in network tools such as apps, emphasizing that AI voice outbound calls have become one of the main means for fraud to reach victims
• On July 1, 2026, Minhou Public Security in Fuzhou issued an emergency warning targeting AI voiceprint cloning fraud, notifying cases where lawbreakers used AI to clone relatives' voices to commit phone fraud, reminding residents to verify relatives' incoming calls through multiple channels
• In 2026, the National Anti-Fraud Center reminded users to turn on incoming call warning features, strengthen technical interception against new fraud methods such as AI voice outbound calls and number spoofing, and simultaneously urged the public not to enable screen-sharing or provide verification codes

How to Protect Yourself

• ✅ When receiving a call claiming to be platform customer service or a claims specialist, hang up immediately and independently dial the official customer service hotline published by the platform for verification. Do not directly call back using the number provided by the other party.
• ✅ Never provide sensitive information such as SMS verification codes, bank card passwords, or payment passwords to anyone over the phone; any customer service behavior claiming that a verification code is needed to complete a refund or unfreeze is a scam.
• ✅ Refuse to download meeting or remote assistance apps from unknown sources, and do not enable screen-sharing functions during a call; if already induced to enable them, disconnect immediately and change relevant account passwords.
• ✅ Install and enable the incoming call warning function of the National Anti-Fraud Center app on your mobile phone to timely identify flagged scam numbers and suspected AI outbound calls.
• ✅ Agree on an exclusive secret code or verification question among family members. When receiving an urgent distress call claiming to be a relative asking for a transfer, use the secret code to verify identity to prevent AI voiceprint cloning impersonation.