Welcome to the AI research bites. This series of short and informative talks showcases cutting-edge research work from ServiceNow AI Research team. The AI Research Bites are open to all, especially those interested in keeping up with the fast-paced AI research community. So, you built your agent and you're ready to deploy—but how do you know it won't be hijacked by malicious users or compromised environments? Enter DoomArena, a comprehensive security testing framework that plugs into real-world agent deployments to reveal vulnerabilities before they become costly breaches. Léo Boisvert shows that testing on state-of-the-art agents reveals surprising security gaps that traditional guardrails can't protect against, giving businesses the critical insights needed to secure their AI systems before potential reputational and financial damage occurs. Paper: https://arxiv.org/abs/2504.14064 ServiceNow AI Research team: https://www.servicenow.com/research/

 171      2