What Is Changing — The AI Basic Act and Data Industry Act Amendments Approved by Cabinet on October 13
Korea is moving to require that AI training data built with government money be shared on a public platform. According to the Ministry of Science and ICT, the amendments approved by the Cabinet on October 13 come down to the following:
Note that this is only the government bill clearing the Cabinet. It still has to go through the National Assembly. Press reports put the effective date at six months after promulgation, but the text and supplementary provisions are not final and could change during legislative review.
Why Smart Factory Companies Should Care
Government support programs are shifting from funding equipment to funding *data*:
The catch is that factory data *is* process know-how. Injection temperatures and pressures, defect images, and vibration or current readings from equipment can tell a competitor exactly how you raised your yield. Deciding what counts as "training data produced through a government project" is likely to become the biggest practical issue.
The scope of covered projects and the procedure for deciding exceptions will be set later in enforcement decrees and ministerial notices, and none of that is final yet. Still, waiting is risky, because the evidence you need to claim an exception is hard to put together after the fact.
Preparing to Use the Exception — Meeting Trade-Secret Requirements
Confidentiality management is the key
Under Korea's Unfair Competition Prevention Act, information is protected as a trade secret only if it is not publicly known, has independent economic value, and is managed as a secret. If your data sits in shared folders without access controls, confidentiality pledges, or labeling, it will be hard to argue later that it can't be shared because it is a trade secret.
Build a data catalog
Track personal data separately
For data that includes personal information, such as worker video or biometric signals from wearables, keep a separate record of whether it was de-identified and which exception applies to it apart from trade secrets.
Revisit Your Contracts With Suppliers and Lead Agencies
Disclosure Is Not Just a Loss — Treating Data as an Asset
Sharing non-sensitive data on AI Hub can get you quality-control support and a place in a shared training-data ecosystem. With manufacturing data productization support, you can also decide which data to sell and which to protect.
Checklist for now
How KITIM Can Help
KITIM helps companies write the data management plans required in manufacturing-AI smart factory proposals, identify what to check in supplier contracts, and design projects that both protect and use their data. If you want to join government programs without giving away your process know-how, contact KITIM for a consultation. We will build a response strategy that fits your company's data.