🧔♂️ A friendly human may check it before it goes live. More news here
South Korea AI anti-phishing platform saves $13m
South Korea’s financial regulator reported that an AI-based anti-phishing platform has prevented about 19 billion won (US$13.32 million) in losses over the past three months.
Launched in late October, the AI-based Phishing Sharing and Analysis Platform (ASAP), enables 130 financial institutions to share information on voice phishing scams, including victim accounts and suspicious transactions.
Between October 29 and January 21, about 148,000 cases of phishing-related data were shared, resulting in the blocking of transfers from 2,705 accounts.
The Financial Services Commission (FSC) emphasized the platform’s role in enhancing digital financial security and rebuilding trust in the sector.
🔗 Source: Yonhap
🧠 Food for thought
Implications, context, and why it matters.
The 19 billion won prevention claim needs clearer attribution and methodology
- Context is missing around the 19 billion won estimate of losses prevented, so ASAP’s real-world effect cannot be judged.
- The way the “prevented” total was computed needs to be spelled out by the Financial Services Commission (FSC), South Korea’s financial regulator. It is unclear whether it counts transfers stopped across 2,705 accounts from Oct. 29 to Jan. 21 or a wider model-based estimate.
- Connection requirements for the 130 participating financial firms should be published for both options. One option is direct operation, the other uses application programming interface (API) calls, so setup effort can be gauged 1.
ASAP may create integration opportunities for vendors, but that is not stated by regulators
- Regulatory technology (regtech) firms, companies that sell compliance and risk-management software, plus payment processors could package integration help for the 130 participating financial firms. Regulators have not said they plan this.
- Providers might use the platform’s API to route ASAP results into internal systems, such as flagging suspected voice phishing and speeding responses. The platform has not been described as supporting automatic blocking through the API.
- Smaller institutions with limited in-house AI may gain the most because participants can choose between direct operation and API calls based on their AI infrastructure conditions 1.
- Participating institutions are expected to receive AI analysis information in the first half of 2026, not 2025 1.
