Zscaler SPLX AI Runtime Threat Inspection is an AI security solution that analyzes LLM interaction logs to identify security and safety threats in GenAI applications. The product performs automated risk triage by scanning uploaded JSON log files against 25+ AI threat detectors to uncover attempted and successful attacks. The solution provides visibility into user prompts and agentic AI workflows to detect harmful inputs and security breaches. It leverages red teaming intelligence to minimize false positives and deliver accurate threat identification. Users can drill down into specific interactions to understand the risk surface of their AI systems and isolate attacks. The platform supports log upload functionality where users select preferred risk scanners and upload LLM logs for instant threat discovery. It identifies various AI-specific threats including prompt injection attempts, security measure evasion, and malicious user inputs across conversational platforms and LLM-based systems. The tool connects to AI systems through REST API integrations, conversational platforms, and various large language models. It provides near real-time insights into AI security posture and enables security teams to remediate critical threats before escalation. The solution is designed for organizations deploying GenAI applications that need continuous monitoring and threat detection capabilities for their AI infrastructure.