AI-powered scam calls exposed – 100,000 a day

A disturbing trend of AI-powered scam calls was brought to light at CCTV's 315 Gala, China's premier consumer rights protection event, held Saturday night.

The expose revealed that software companies are developing bot tools capable of making "100,000 calls daily, tirelessly."

Companies like Shanghai-based Zhiyouqing Network Technology, Shanghai-based Feige and Zhongshan-based Lvxin have been leveraging artificial intelligence to mimic human voices and bypass regulatory oversight, triggering nationwide concern.

These robocalls deliver promotional messages for services like education, dental care, and fitness memberships, which are widely considered intrusive.

A CCTV report featured a Zhiyouqing manager demonstrating how their "robots" utilize AI-generated human voices and keyword-triggered responses to simulate natural conversations. The AI can even adapt dialogues based on consumer reactions, switching tactics when a user expresses disinterest.

These AI-driven systems operate around the clock, making up to 100,000 calls daily. The ability to convincingly mimic human voices makes it difficult for consumers to distinguish between AI and genuine human callers, increasing answer rates, according to a manager named Tian in the CCTV report.

To evade legal scrutiny, these companies exploit "virtual numbers" (16/17-prefixed numbers), which don't require SIM card or identification authentication, making them difficult to regulate. Additionally, location-masking technology allows these calls to appear local, further enhancing pick-up rates.

"We sign dozens of clients monthly," boasted Tian, whose company services more than 5,000 clients across sectors like health care and finance.

Following the expose, Shanghai market regulators raided Zhiyouqing's offices late Saturday night. According to the Qixinbao, an enterprise database of Hehe, Shanghai Zhiyouqing was founded in September 2021 with a registered capital of 1 million yuan (US$138,888).

Telecommunications officials from China Mobile and China Telecom acknowledged the difficulty in cracking down on these activities, as the virtual numbers are generated by AI rather than traditional carriers.

Deepfake and a black industry chain

This issue highlights the growing problem of deepfakes, which use realistic impersonations to deceive people into divulging personal information or transferring funds.

During the recent Two Sessions, political advisers, including Xiaomi CEO Lei Jun, expressed concern over the increased accessibility of AI tools, the lower technical barriers, and the ease of creating anonymous deepfakes.

Another CCTV report revealed companies developing tools to collect user data from short video platforms like Douyin (China's equivalent of TikTok) without permission.

They then use AI to process and "customize" the data, including location, age, and sensitive information like marital status, pregnancy, and property ownership. This creates a "black industry chain" that makes scam calls and other deepfake frauds more targeted and dangerous.

Experts are calling for stricter regulations over virtual operators and AI ethics frameworks to combat deepfakes.

Security firm Kaspersky advises users to avoid answering 16/17-prefixed numbers and to hang up immediately if they do answer. They also recommend avoiding any interaction with robot-led conversations, including saying "yes".