AI NewsWaveKorea Deep Learning Cuts IBK Document Processing for Voice Phishing Relief by 88%
Key points
- VLM-based voice phishing relief automation demonstrated
- Document processing time cut by 88%
- Excellence Award in the PoC category of the Public Institution AI Innovation Challenge
Korea Deep Learning, a document AI company, said on the 30th that it worked with IBK Industrial Bank of Korea's Financial Consumer Support Department to automate voice phishing damage relief work using a vision-language model (VLM), cutting document processing time by 88%. According to Datanet and Techworld News, the project was led by IBK's Financial Consumer Support Department, with Korea Deep Learning taking part as the core technology provider, and ran from April to August this year.
Platum explained that victims of voice phishing must submit documents such as a relief application and a copy of their ID to the bank, and said in its headline that the process of handling faxed documents by hand was replaced with AI. Platum also reported that the company completed a demonstration of automating this work with an AI agent. Industry News reported that Korea Deep Learning described the case as the financial sector's first VLM-based automation of voice phishing relief work.
News of an award also came out. According to AI Times, the project was selected for the Excellence Award in the AI service proof of concept (PoC) category of the 2026 Public Institution AI Innovation Challenge. Korea NGO Times reported in its headline that IBK received the Deputy Prime Minister's Award, and wrote that IBK was rated highly at a government-organized innovation competition held in Mapo-gu, Seoul, on the 28th. The outlet said it remains to be seen whether shortening the time from receiving a report to providing relief can strengthen financial consumer protection. The name of the award was reported differently by outlet.
Recently, responses to voice phishing and uses of AI have continued across the financial sector. Chosun Biz reported on the 29th that Samsung Life Insurance is expanding consumer protection activities centered on preventing financial fraud, managing complaints and improving customer notices. According to the report, Samsung Life looks for abnormal signs in transactions to prevent damage, and also has verification procedures to prevent voice phishing damage when customers cancel insurance contracts or take out loans. Its headline said difficult notices are made easier with AI.
Glossary A vision-language model (VLM) is an AI model that understands and processes images and text together. A PoC refers to a proof of concept, which verifies on a small scale whether a technology is actually effective before full-scale adoption.