Security Implications of Chinese-Developed LLMs

In late January, the Chinese start-up company DeepSeek introduced an open-source large language model (LLM) that led to billion-dollar impacts on the stock market and myriad questions around the future of LLM development and use. In this webinar, IANS Faculty Summer Fowler and Jake Williams discuss the security risks with LLMs developed by Chinese organizations. While keeping the math to a minimum, some technical background in LLMs is critical to understanding these issues. Expect to learn some about how LLMs actually work, what "open source" actually means for an LLM, and why it's impossible to fully audit a model. We dive into:

• Specific issues and recommendations around everyone's favorite Chinese-developed LLM
• An explanation of what risks can't be mitigated and what can
• Recommendations for understanding and communicating residual risks