Christian Szandor Knapp

With Agentforce, Salesforce has opened the box of wonders for us, and we can augment our workforce with powerful agents that deliver scalable service around the clock. But what looks like a technology miracle for some is yet another attack vector for others. And we all want benefit from the AI revolution without comprimising the security of our data. This session is going to give you an overview about how prompt attacks against GPTs work, how Agentforce's security model and prompt defense is designed and how you can make sure that your agents are helpful and secure instead of chatty.