Losses from direct and secondhand-trading fraud in South Korea reached a record 874.07 billion won last year, according to police data submitted to the National Assembly. Generative AI is adding a new problem by making convincing fake receipts, cargo documents and transaction records easier to create.
The police figures cited in the report also recorded 119,741 cases, showing that the problem is large in both money and volume. The danger is not simply that forged documents look better than before; it is that familiar visual cues can no longer be treated as reliable proof that goods, payment or shipping actually exist.
Why visual proof is getting weaker
A recent Jeju police case illustrates the scale possible with fabricated material. Police said they arrested 31 people connected to fraud groups, including one organization accused of using false product and transaction information to take 2.4 billion won from 1,091 victims.
AI does not create the underlying fraud model, but it lowers the effort needed to produce supporting material that looks plausible at a glance. That weakens older habits such as trusting a screenshot, receipt image or shipping form simply because it appears professionally formatted.
The more durable defense is to rely on transaction systems that verify payment and delivery independently of images supplied by a seller. As generative tools improve, platform controls and traceable payment processes become more important than trying to judge authenticity by appearance alone.