Why LG Is Suing a Junior AI Researcher Over a $8M Job Offer

The $8 Million Job Offer That Triggered a Lawsuit

LG AI Research tried to stop a junior researcher from leaving. The court said no. LG appealed anyway.

The case centers on a researcher known as A — a master’s degree holder with two years of experience at LG’s AI lab, working on post-training for the K-EXAONE 2.0 foundation model under the government’s “Dokpamo” (독파모) project. When A received an offer from UpStage, LG moved fast. They offered A a 15% base salary increase and an additional 20 million won per year for three years to stay. When that didn’t work, LG filed for an injunction.

The first-instance court rejected it. LG is now appealing.

This is not a trivial dispute over a handful of researchers. UpStage reportedly offered A stock options worth up to 800 million won — structured as 4+4, meaning half vested after four years and the other half after eight — if A joined before UpStage’s extraordinary general meeting on April 30. That timing matters. It suggests the hiring was deliberate, not opportunistic.

LG’s Argument: It’s Not About Title, It’s About Access

LG’s position is straightforward and somewhat unusual for a case involving someone labelled “junior.” The company argues that A should not be classified as a typical two-year junior developer. According to LG, A belonged to the top 5% of core researchers at LG AI, received retention bonuses and internal awards reserved for high performers, and consistently received top performance reviews.

More importantly, LG contends that A accumulated irreplaceable R&D assets during post-training — success and failure patterns, trial-and-error insights, internal knowledge that cannot simply be “unlearned” by switching employers. LG frames this not as a case of stopping someone from working elsewhere, but of protecting trade secrets that cost millions to develop.

UpStage pushes back hard. The company says A applied through open recruitment, is handling a different technical domain than LG, and is not being hired for any specific project or technology. UpStage also points out that both A and another departing executive are listed as contributors on the Dokpamo Phase 2 technical reports for both K-EXAONE 2.0 and Solar Open2 — information that UpStage says can be fully disclosed and verified.

What the Courts Are Really Deciding

The legal question here is sharper than it appears on the surface. This is not about whether someone can change jobs. It is about what information a researcher carries with them — and whether that information belongs to the employer.

Korean courts are now being asked to draw a line between protected trade secrets ( proprietary data, unpublished experiment results, internal evaluation materials, specific training methodologies) and the general skills, experience, and knowledge a researcher naturally accumulates. The distinction matters enormously for an industry still figuring out what counts as intellectual property in the age of foundation models.

The trend is familiar. xAI has sued former OpenAI employees over alleged misappropriation of confidential information. Apple and OpenAI have clashed over the same terrain. In Korea, Naver sent a formal notice to SK Telecom in 2023 over an AI specialist’s departure — though it never escalated to litigation.

What makes LG’s case distinct is the seniority — or lack thereof — of the subject. Suing a researcher with only two years of experience sets a tone. It signals that even mid-level talent is now considered too valuable to let move freely.

The Ecosystem Cost

Industry insiders are watching closely. Several AI companies told E-Daily that using injunctions as a broad talent-retention tool would be a mistake. One pointed out that companies like Rebelion, FuryoAI, and UpStage — all backed by the National Growth Fund — are competing for the same pool of talent within the same ecosystem. Natural movement between them is part of how the industry grows.

“Restricting personnel movement too broadly would shrink the entire Korean AI ecosystem,” said one AI technology company representative. “If you want to keep core talent, improve compensation to match market value — don’t sue them into staying.”

Lee Sung-yup, director of the Korea University Technology Law and Policy Center, agreed: “Researcher mobility and knowledge diffusion are vital to national AI competitiveness. Injunctions should be used extremely restrictively.”

The concern is real. Korea’s government is pouring unprecedented public funds and computing resources into the Dokpamo project. Companies are racing to capture talent before the window closes. But if every departure becomes a courtroom battle, the net effect may be less innovation, not more — researchers will think twice before joining startups, and startups will struggle to staff up at all.

Who Wins, Who Loses

Right now, LG lost the first round. But the appeal changes the calculus. A prolonged legal fight — even if LG ultimately prevails — has a chilling effect on everyone. Junior researchers at Korean AI labs will read this case and reconsider their options. Startups like UpStage will factor litigation risk into hiring decisions. The cost of moving between companies just went up, even for people who did nothing wrong.

The real loser may be Korea’s broader AI ambition. The country needs mobile, collaborating talent to compete with US and Chinese firms. Locking researchers into employers through legal threats is a short-term fix with long-term consequences.

The case will likely establish a precedent for how Korean courts treat trade secrets in AI research — specifically, how they distinguish between protectable proprietary information and the general expertise a researcher carries. That precedent will shape the industry for years.

For now, the courts have drawn a line: an injunction is not a standard talent-retention tool. LG is appealing. The question is whether the next court agrees.