Zscaler SPLX Automated AI Red Teaming is a platform that performs automated vulnerability assessments and simulates attacks on AI systems throughout their lifecycle from development to production. The platform tests AI applications against various threat categories including prompt injections, hallucinations, social engineering, and off-topic responses. The system provides 25+ prebuilt probes for testing AI security and safety risks, with the ability to create custom probes for domain-specific testing scenarios. Users can upload custom datasets of attack prompts tailored to their threat models and run targeted evaluations. The platform supports multi-modal testing across text, voice, images, and documents to assess AI assistants with different input types. It simulates attacks from both adversarial and regular user personas to uncover vulnerabilities in different interaction scenarios. Testing can be integrated into CI/CD pipelines for continuous security assessment throughout the AI application lifecycle. The platform evaluates existing AI filter mechanisms and security guardrails to refine them for optimal protection. Results include automated compliance mapping to frameworks such as MITRE ATLAS, NIST AI RMF, OWASP LLM Top 10, Google SAIF, EU AI Act, ISO 42001, DORA, and Databricks DASF. The platform provides remediation guidance based on discovered risks and supports issue tracking integration. The platform connects to AI systems through REST API integrations, conversational platforms, and large language models without requiring coding.