In 2024, we hacked Googles A.I Gemini and leaked its source code, earning the Most Valuable Hacker title at the bugSWAT event in Las Vegas. The world of Generative Artificial Intelligence and Large Language Models is evolving rapidly, with tech giants like Meta, Microsoft, and Google competing for dominance. As AI assistants and classifiers become more prevalent, the importance of building securely is paramount. AI agents are revolutionizing decision-making processes with chain-of-thought reasoning, leading to more adaptive and reliable systems. Googles proactive approach to AI security, exemplified by events like LLM bugSWAT, demonstrates their commitment to identifying and addressing vulnerabilities. Our discovery of a new vulnerability in Gemini at the bugSWAT event showcases the ongoing evolution of AI security.
