Naman Mishra, CTO of Repello AI and an expert in AI security, dives into the complexities of securing large language models. He discusses layered vulnerabilities and highlights alarming risks like prompt injection and data leaks, including a fascinating anecdote about a Windows activation key leaked by ChatGPT. Naman emphasizes the importance of continuous red teaming as a proactive approach to security, and explores the crucial role of ethical hackers. The conversation sheds light on the urgent need for robust security measures in AI technologies, especially in sensitive sectors.
