Hyundai Motor blocks external generative AI in core research units such as autonomous driving
Hyundai Motor has cut off the use of external generative AI in its core research units. According to industry sources on the 4th, Hyundai Motor Group has restricted the use of external generative AI in core research and development organizations—such as autonomous driving and robotics—since late last month. The move is intended to prevent source code (a program's original blueprint) and design information from leaking to external servers. As a result, Hyundai has excluded not only external generative AI services such as Claude, but also Chinese-developed AI models like DeepSeek and Qwen from approved usage.
External generative AI services are structured so that users' questions or source code are transmitted to the service provider's servers. Concerns have long been raised that undisclosed design information and research data could remain on external systems. If a researcher enters undisclosed design information or research data while verifying code or writing documents, that information could be left outside the company.
In particular, some of the technologies Hyundai handles, including autonomous driving and hydrogen, are designated as national core technologies. National core technologies are separately managed under the Industrial Technology Protection Act because their overseas leakage could have serious consequences for national security and the economy. Autonomous driving and robotics involve large volumes of sensitive research information, including source code, driving data, and sensor data.
Hyundai has not blocked AI use altogether. Only AI services that transmit data to external servers are restricted, while AI running on in-house infrastructure remains available. This approach blocks the possibility of research data being exported externally while continuing AI use in development settings. The company's internal generative AI platform 'H-Chat,' introduced in 2023, is used by more than 70 percent of its employees. The 'Crash Safety AI Assistant' at the Namyang R&D Center consolidates searches of test and analysis materials, cutting engineers' document search time by roughly 90 percent.
Similar moves are appearing at other major companies. Samsung Electronics restricted the use of external generative AI via company PCs in 2023 after it came to light that some employees had entered semiconductor-related source code into ChatGPT. SK hynix also blocks access to ChatGPT on its internal network as a rule, allowing use only in cases approved after a security review. POSCO likewise restricts generative AI use to its internal intranet.
