62 views
It has now been 2 years since the first draft of the OWASP Top-10 for Large Language Model Applications was published and brought topics such as prompt injection, data poisoning and much more into focus. Most companies - and attackers - now have GenAI-based solutions in production, and potential problems have become real risks, and every week new topics, risks and defense strategies are presented, including agents that implement exploits faster than the operations team can react. We present you the top 10 biggest vulnerabilities of LLMs and show you how to deal with the risks. By the way: You can download the OWASP Top-10 for free in German at the following URL: https://mayflower.de/ki-navigator-tal...