A career spent tracking illegal activity online has taken on a new purpose in the age of artificial intelligence (AI).

For nearly two decades, Choi Seung-hyun worked in software license compliance and anti-piracy, with stints at Adobe, Microsoft, the Business Software Alliance and PTC. Today, as CEO of DragonEyes, he is applying that experience to a different problem: an AI agent designed to detect harmful online content and help protect children.

"My 20 years in software license compliance were a major inspiration for DragonEyes," Choi told The Korea Times in an interview. "It started with a question i had as a compliance professional about how we could tackle harmful content online. That eventually led me to develop an AI agent."

The idea became personal in November 2025, when Choi read about a multinational game company in the United States facing investigations over its handling of harmful online content. Some of the cases had raised concerns about child abuse. Choi, whose two sons were then in high school and elementary school, began examining the issue both as a parent and as a compliance expert.

"As a father, I couldn't help but pay attention," he said. "I started looking closely at those cases from a compliance perspective, and I think that was when I began thinking seriously about harmful online content."

The timing reflects a broader moment for the industry. Platforms and game companies around the world have faced growing scrutiny from governments and regulators over how they handle harmful content, including material that endangers children, and pressure has mounted for faster and more consistent detection and removal. AI-assisted moderation has become a common answer to the sheer volume of content, though debates continue over accuracy, accountability and the role of human judgment.

After three months of planning, Choi began developing DragonEyes in February and unveiled it six months later, on Aug. 25, at a National Assembly seminar. The choice of venue points to where much of the discussion around online safety is now taking place: with lawmakers, as countries including South Korea weigh how to strengthen obligations on platforms to protect minors.

DragonEyes uses AI to search for and analyze potentially harmful content, but the final decision rests with a human reviewer. That human-in-the-loop design mirrors a widely held principle in content moderation, where automated systems flag potential violations at scale while people make the final calls on ambiguous cases. Choi, who studied law rather than engineering, said the structure reflects his view of technology.

"Law, society and technology should all ultimately put people at the center," he said. "The purpose of DragonEyes is, in the end, to protect people and make their lives better."

That human role led Choi to consider turning content review into jobs for people with disabilities and older adults. This was not part of his original plan. In the early stages, he wondered who could perform the reviewing and how the work could be funded.

Conversations with Kwon Chan, a former Microsoft colleague and former vice president of ChorogUsan for Children, had also made him aware of the shortage of stable, accessible jobs for people with disabilities. Employment for people with disabilities remains a persistent challenge in South Korea, where the disability employment rate has long lagged that of the general working-age population.

Choi is reluctant to dress the idea up in grand terms.

"'Digital humanism' sounds a little too grand for me," he said. "I think of AI more as a reliable friend — something that helps us with the things that are inconvenient or tedious to do alone."

His hope for DragonEyes is equally straightforward.

"DragonEyes could become a valuable job for someone, and for someone else, a way to connect with the world," he said.