Indonesia's financial regulator has blocked Rp674.10 billion in fraud proceeds since its anti-scam centre launched 19 months ago, but only Rp196.93 billion, about 29 percent, has been returned to victims. The Rp477.17 billion still in processing shows where the system struggles: freezing accounts can happen quickly; recovering money that has already changed hands is a different problem.

OJK (Indonesia's Financial Services Authority) Board of Commissioners Chair Friderica Widyasari Dewi presented the figures at an anti-scam defense event in Jakarta on Monday, July 6, 2026. She warned that the latest threat comes from fraudsters adopting artificial intelligence and deepfake technology. "Advances in artificial intelligence and deepfake technology make it easier for perpetrators to impersonate someone the victim knows," she said.

IASC (Indonesia Anti-Scam Centre) is a coordination body established by OJK alongside the financial services industry, law enforcement, and payment system operators to cut off fraud fund flows through rapid account blocking. From its launch in November 2024 through June 24, 2026, the centre received 608,168 fraud reports and blocked 557,751 accounts.

Why is victims' money so hard to recover?

Fraud proceeds leave a victim's account in under an hour on average. That speed is why the 29 percent recovery rate is so hard to push higher: once money passes through several layers of nominee accounts, payment channels, or crosses into international networks, IASC loses control of the funds even when blocking is still possible.

Each transfer layer adds time and jurisdictions to resolve before money can be returned. Perpetrators build pathways through courier accounts, merchants and sub-merchants, and virtual assets with international reach.

From hacking to psychological manipulation

Perpetrators now manipulate victims into voluntarily handing over one-time passwords, credentials, or transferring funds directly. Bank security systems remain technically intact; it is the victim's psychology that is exploited.

Deepfake technology makes this worse. Friderica described how it dismantles the recognition mechanism people have relied on: a person can now receive a video message from a face and voice they know, when the entire exchange has been digitally fabricated using AI.

UNODC data cited by OJK puts one in four Indonesians as a victim of financial fraud. Friderica said no segment is immune, with cases spanning all education levels and backgrounds.

Education racing against new tactics

Friderica acknowledged one structural limitation openly: by the time a fraud method is publicized, perpetrators have already moved to a new one the public does not yet know about.

Public awareness campaigns always react to methods that already exist, while perpetrators keep developing new ones. The race is never even.

Friderica also said the 608,168 report figure understates the true scale. "I believe this is only the tip of the iceberg, because not everyone will report that they have become a fraud victim," she said. If her estimate is correct, actual losses far exceed the Rp674.10 billion IASC has detected.

That burden falls on consumers whose financial room is already tight. Alongside inflation at 3.34 percent still driven by rice and food prices, and fiscal pressure from a sharply widening budget deficit, unrecovered fraud losses add another layer of economic pressure on households.

OJK's response and two key numbers

OJK has prepared four response tracks: strengthening risk-based anti-money-laundering governance, improving customer due diligence including digital identity verification, developing AI-based detection and behavioral analysis systems to flag suspicious transactions in real time, and expanding financial literacy.

Two numbers will show whether the response is enough. First, the recovery ratio: will the Rp477.17 billion still in processing push returned-to-victim funds past 29 percent, or will the figure stall? Second, whether OJK will issue technical anti-deepfake standards for banks and fintech firms, such as mandatory liveness detection, so that protection does not rely solely on public literacy racing against methods that keep changing.