Technical Architects, Stay Ahead of the Top 6 GenAI Security Risks
Gartner senior director analyst, Dennis Xu, went into deep detail about generative AI apps and what we should watch out for moving forward. He explained that data can be leaked so improving security is very important. He talked about things like direct and indirect prompt injection and how wording a question can produce such different outcomes. For example, if you ask ChatGPT “how to make a bomb?” It will not give a response, but if you rephrase it then it will give a response. He also mentioned other methods like hallucination meaning that AI knows a lot of information but it can not understand information like a human can. I liked learning about this topic because I never knew security had to be implemented for generative AI apps.