AI-Manipulated Images: The New Face of Refund Fraud in China

AI-Manipulated Images: The New Face of Refund Fraud in China

Admittedly, I splurged quite a bit on online shopping this holiday season. Naturally, not all purchases lived up to my expectations. For instance, a photobook I ordered arrived damaged. I simply captured some images of it, emailed them to the retailer, and secured a refund. Online retailers have traditionally relied on customer-submitted photos to verify legitimate refund requests. However, this long-standing method is now facing disruption.

Signs of Deception

On the Chinese internet, WIRED found several complaints from ecommerce merchants and customer service workers about dubious refund requests backed by what appear to be AI-generated images. In one instance, a customer claimed that their bed linen set arrived in shreds, yet the Chinese text on the label looked nonsensical. Another buyer presented a photo of a coffee cup with what seemed like paper tears, despite it being ceramic. 'How can you tear ceramic like paper?' the seller puzzled.

Sellers noted that particular types of products are especially vulnerable to AI-aided photo fraud: perishables, luxury items, and fragile goods like ceramics. Typically, merchants do not require returns of such items before issuing refunds, creating an opening for fraudulent returns.

In an incident on Douyin, China's version of TikTok, a live crab merchant received a complaint accompanied by a photo suggesting that the crabs arrived dead, with some even 'escaping'. Videos allegedly show a finger prodding the lifeless crabs, but anomalies were evident.

Gao Jing, who comes from a family with over 30 years of crab farming experience, was skeptical. 'Dead crabs with legs pointing upward is something we've never seen,' Gao explained in a Douyin video post. The ruse was revealed by the gender inconsistency of the crabs across two videos and one crab displaying more legs than normal.

Gao reported the fraud to authorities, and following investigation, the buyer was detained for eight days as announced by a shared police statement online. The occurrence caught widespread attention on Chinese social media, perhaps because it marked the first AI-driven refund scam to lead to legal consequences.

The Rise of Technological Fraud

This issue is not confined to China. Forter, which specializes in fraud detection, states that globally, AI-altered images in refund claims have risen by over 15% this year.

As noted by Michael Reitblat, Forter's CEO and cofounder, the trend gained momentum mid-2024 but has accelerated due to the broad availability and ease of use of image-generating tools. Retail staff might lack the resources to thoroughly vet every suspected image forgery.

Reitblat also indicates that criminal groups are using similar approaches for large-scale refund deception. For example, scammers filed over a million dollars in false refund requests utilizing image manipulation to simulate damage to household products.

Retailers are also experimenting with AI to detect doctored images. One Chinese toy seller shared with WIRED that they use AI chatbots to evaluate refund requests for forged images, although these systems aren’t foolproof yet.

Despite this technological duel, prevailing refund policies might not favor sellers solely based on AI indications. Reitblat cautions that e-commerce companies could tighten return policies, potentially penalizing honest consumers.

This situation mirrors past criticisms of sellers employing AI to create misleading product images. Buyers complained that online shopping had turned into a gamble, uncertain if the actual product would match its representation.

Ultimately, these practices underscore a core issue: e-commerce's dependency on trust is strained by unabated AI accessibility. The platform might need to implement new verification processes, adjust refund protocols, or enhance accountability measures to manage AI-facilitated fraud without compromising customer faith.

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Posts