Research

Kolmar Korea bets on AI to predict skin safety before cosmetics are made

Kolmar Korea signed an MOU with the Korea Research Institute of Chemical Technology to build AI models predicting skin reactions and ingredient absorption before cosmetics prototypes are made.

Kolmar Korea turns to AI to predict skin safety before cosmetics hit market
Kolmar Korea turns to AI to predict skin safety before cosmetics hit marketKost9n4 / pixabay
By Elena Vasquez5 min read

Why it matters

  • Kolmar Korea signed an MOU with the Korea Research Institute of Chemical Technology on Tuesday, announced Wednesday.
  • Moon Byung-seok of Kolmar Holdings and Shin Seok-min of KRICT signed the agreement.
  • The AI models will predict skin reactions and ingredient absorption for both individual ingredients and multi-component formulations.
  • Target categories include organic sunscreen agents, preservatives and surfactants subject to strict country-by-country regulation.
  • Kolmar Korea expects the system to cut development time and costs and support overseas market preparation.

Kolmar Korea signed a memorandum of understanding on Tuesday with the Korea Research Institute of Chemical Technology to develop AI technology that predicts the skin effects of cosmetic ingredients and formulations before any product is manufactured. The company announced the deal Wednesday, framing it as a way to speed up development and strengthen the global competitiveness of its clients.

The signing took place between two named executives. Moon Byung-seok, head of technology research at Kolmar Holdings, signed for the corporate side. Shin Seok-min, president of the Korea Research Institute of Chemical Technology, signed for the government research institute.

The stakes here are practical, not abstract. Cosmetics manufacturers routinely sink time and money into prototypes that later fail safety evaluations, and formulations built for one market can stumble on another country's rules. A prediction system that runs before the first prototype exists would change where in the pipeline that risk gets resolved.

What will the AI models actually predict?

The two organizations will collaborate on research, technical exchanges and performance assessments to build AI models capable of predicting two things: skin reactions and ingredient absorption.

That second element matters as much as the first. Absorption determines how much of a substance passes into the skin, which is central to how regulators judge whether an ingredient is safe at a given concentration. A model that estimates absorption alongside irritation risk gives formulators a fuller safety picture.

The technology is designed to assess not only individual ingredients but also complete formulations containing multiple components. That distinction is significant. Real cosmetics are mixtures, and mixtures can behave differently than their parts. According to the announcement, using ingredient combinations and formulation data, researchers will be able to identify potential safety concerns earlier in the development process than current practice allows.

Who brings what to the partnership?

The division of labor is explicit in the agreement.

  • The Korea Research Institute of Chemical Technology contributes its accumulated expertise in AI-based toxicity prediction, built up for both individual substances and chemical mixtures.
  • Kolmar Korea supplies actual ingredient and product data, which the company will use to assess the technology's commercial potential.

That pairing addresses the two chronic bottlenecks in applied AI for chemistry: credible predictive science on one side, and real industrial data to validate it on the other. Institutes rarely hold proprietary formulation libraries. Manufacturers rarely hold toxicity-prediction research programs. The MOU is an attempt to close that gap for the cosmetics sector specifically.

Which ingredients are in the regulatory crosshairs?

The system is expected to be particularly useful for formulations containing ingredients subject to strict regulations in different countries. The announcement names three categories:

  • Organic sunscreen agents
  • Preservatives
  • Surfactants

These are among the most heavily scrutinized substance classes in global cosmetics regulation, and rules for them vary across jurisdictions. Sunscreen filters in particular are treated as borderline products in some markets and as cosmetics in others, with distinct approval pathways. A prediction tool that flags regulatory exposure early would let manufacturers adjust formulations before committing to production runs.

Kolmar Korea expects the technology to help cosmetics manufacturers anticipate international safety requirements and prepare products for overseas markets. For a contract manufacturer whose business depends on shipping clients' products across borders, that is a direct commercial capability, not a lab curiosity.

What changes in the development pipeline?

The current sequence in cosmetics development is straightforward: formulate, prototype, test, iterate. Safety problems surface at the testing stage, after materials have been purchased and prototypes produced. Each failed iteration costs time and money.

The planned system moves that evaluation upstream. Researchers would query AI models with ingredient combinations and formulation data before producing anything physical. Potential skin effects would be estimated digitally, and only promising candidates would proceed to prototypes.

Once development is complete, Kolmar Korea plans to incorporate the system into its research and development process. The company says this could reduce the time and costs involved in developing products for its clients. No timeline for completion was disclosed in the announcement, and no quantitative targets for cost or time savings were given.

What did Kolmar Korea say?

The company put its ambitions in the context of K-beauty's global position. "Through this agreement, we have established a foundation for assessing skin safety from the earliest stages of product development," a Kolmar Korea official said.

The official added: "We will actively support our clients' expansion into global markets while developing K-beauty products that consumers around the world can trust."

The second quote connects the partnership to the wider K-beauty export drive. Korean cosmetics have built global market share on speed to market and trend responsiveness, and the official's framing positions early safety prediction as infrastructure for sustaining that pace while meeting divergent international standards.

Why does an MOU between a manufacturer and a state institute matter?

Memoranda of understanding are starting points, not finished products, and this one leaves key questions open — development timelines, performance benchmarks, and whether the models will be offered beyond Kolmar Korea's own R&D pipeline. But the structure of the deal signals where applied AI in cosmetics is heading: toward predictive toxicology as a standard step in formulation, rather than an afterthought.

For the Korea Research Institute of Chemical Technology, the partnership is a route to industrial validation of its AI toxicity-prediction work on real product data. For Kolmar Korea, it is a bet that screening formulations computationally will compound into faster, cheaper development cycles for its clients. If the completed system performs as intended, the next signal to watch is its integration into Kolmar Korea's R&D process and any reported reductions in development time for regulated product categories like sunscreens.

Source: feed.koreatimes.co.kr

Share this article:

More from Elena Vasquez

Elena Vasquez

Show full bio

Market editor covering technology and product news in artificial intelligence and machine learning at AI In Context.

708 articles

Related articles

  1. MIT, BASF test CheMeleon for low-data molecular prediction
  2. Moonshot AI redirects Kimi from coding toward scientific research
  3. China's Drugmakers Turn to AI Marketing as Anti-Graft Rules Bite
  4. Google Publishes Co-Scientist in Nature, Opens AI Hypothesis Tool to Researchers
  5. Over 300 Delegates From 40 Countries Gather in Seoul for THE Innovation & Impact Summit 2026

« Previous articleNext article »