South Korea’s New AI Law Reaches Hiring Tools Built Anywhere, Including India

South Korea’s AI Basic Act took effect on January 22, 2026, and buried in its scope is a line that should matter to any India-based team screening candidates for a Korean employer: the law’s extraterritorial reach extends to activities outside the country that affect the Korean market or its users. In practice, that means an AI resume-screening or interview tool used to evaluate Korean employees or job applicants can trigger obligations under the Act regardless of where the tool was developed, hosted, or operated. Build it in Bangalore, run it on servers in Mumbai, screen candidates for a Seoul-headquartered client, and the law still reaches you.

What actually counts as “high-impact” here

The Act’s “high-impact AI” category is defined by function, not geography: systems used for judgments or evaluations that affect a person’s rights or obligations, and employment-related determinations sit squarely inside that definition. Resume screening, candidate ranking systems, performance-evaluation algorithms, and promotion or compensation decision tools are all named as the kind of HR AI use case the statute is aimed at. That’s a wider net than it first sounds, because it doesn’t ask whether the tool is marketed as “AI hiring software.” It asks whether the tool’s output changes what happens to a candidate’s or employee’s application, interview outcome, or standing, which is a functional test most screening pipelines would fail to dodge even if they wanted to.

The obligations that come with that classification aren’t exotic by 2026 standards: notify people in advance that AI is being used in the decision, maintain risk-management documentation across the system’s lifecycle, and keep a human oversight mechanism in the loop rather than letting the model’s output flow straight through unreviewed. What’s less familiar is the foreign-operator trigger. Multinational employers cross into direct obligations, including designating a domestic Korean representative, once they hit thresholds like 1 trillion won in global revenue, 10 billion won in Korean domestic sales, or 1 million daily Korean users. A large GCC parent or a Korean conglomerate’s India delivery center can sit well under those thresholds individually while the parent entity clears them easily, and the parent’s compliance posture is what governs the group’s hiring AI use.

Why this is an India problem, not just a Korea problem

India runs a meaningful share of the world’s HR technology delivery for exactly this kind of multinational: GCCs for Korean electronics, auto, and steel conglomerates operate recruiting and screening functions out of Bangalore, Gurugram, and Hyderabad, and Indian HR-tech vendors sell screening tools into Korean-headquartered clients directly. None of that changes under the Act’s extraterritorial language, because the statute doesn’t care where the vendor or delivery team sits. It cares whether the output of the tool affects a Korean employee or applicant’s outcome. A screening pipeline built entirely by an Indian team, running entirely on Indian infrastructure, for entirely legitimate business reasons, is still in scope the moment it makes employment-affecting judgments about people the law protects.

We’ve made this same argument about the EU AI Act’s reach into India-run screening pipelines: jurisdiction in AI regulation increasingly follows the candidate, not the vendor’s zip code. Korea’s Act is the newest version of a pattern that’s now shown up across the EU, the UK, and a growing list of U.S. states. Where a screening tool is written stopped being the compliance-relevant question years ago.

The honest complication

Enforcement detail here is genuinely thin right now. Early guidance suggests initial enforcement will lean toward regulatory guidance and ecosystem-building rather than immediate penalties, and several implementation specifics are still waiting on presidential decrees rather than final rules. That’s a real reason not to panic-rewrite a screening pipeline this quarter. It’s not a reason to ignore the Act, because “guidance-first enforcement” historically has a runway of months, not years, before regulators start treating early guidance as the baseline they audit against. Teams that build the notice and human-oversight habits now aren’t racing a deadline later.

The take

If your screening pipeline touches a Korean employer’s candidates, either directly as a GCC or indirectly as a vendor, the question worth asking this quarter isn’t “does Korean law apply to us in India.” It almost certainly does the moment your tool makes an employment-affecting call about a person the Act protects. The question worth asking is whether your current process gives candidates advance notice that AI is involved, and whether a human reviewing an AI’s shortlist actually has the standing and time to reverse a call, not just the title of reviewer.

If you’re running screening for Korean-facing roles out of an India delivery center and want to check whether your review step and disclosure language would hold up, talk to us about what an audit-ready screening record looks like.

Sources: Understanding South Korea’s New AI Law: Key Considerations for Multinational Employers — Littler, South Korea AI Basic Act: What Foreign Companies Must Know in 2026 — Safe AI for Business