Non-listed firms take up 70 pc of financial AI patent application in S. Korea

Seoul, Feb 17 (IANS) Nearly 7 out of 10 patent applications filed in South Korea for investment-related artificial intelligence (AI) came from non-listed companies and startups, a report showed on Tuesday, suggesting venture firms are leading the AI transformation in the finance sector.

According to a report from the Korea Capital Market Institute, 67 percent of patent applications for financial investment-related AI were filed by local Software as a Service, or SaaS, companies, reports Yonhap news agency.

Of them, 76 per cent were designed to be sold to other companies or businesses, while only 14 percent were intended for individual customers, according to the report.

Many of the patent requests were related to AI designed for standardised tasks, including consigned trading of financial products, monitoring of abnormal trading activity and stock price analysis.

AI patent filings were minimal in areas that treat non-disclosed information and depend on human networks, such as deal sourcing, as well as areas accompanied by high financial risks, the report said.

“For AI innovation to spread across the entire financial investment industry, rather than being concentrated in certain areas, the active roles of the government and regulatory authorities are necessary,” it added.

Meanwhile, South Korea’s major brokerage houses racked up record earnings last year on the back of a bull run on the local stock market, industry data showed.

The combined net profit of the country’s 27 securities firms is estimated at 10.23 trillion won ($7.03 billion) for 2025, sharply up from the previous year’s 6.97 trillion won.

Korea Investment & Securities Co. posted the biggest net income of 2.01 trillion won last year, up 79 percent from a year earlier, followed by Mirae Asset Securities Co. with 1.59 trillion won in net profit.

The average daily turnover of the local stock market stood at 16.9 trillion won in 2025, marking a whopping 57 percent spike from a year before, according to the data.

—IANS

na/