Knowledge Communication

Generated AI-security Solutions

Knowledge Communication

☆

Not rated

To address risks specific to generative AI, such as prompt injection, PII leakage, hallucination, and model poisoning, we design and implement a systematic guardrail system from four perspectives: input control, output auditing, training data management, and continuous improvement. This builds a secure generative AI utilization platform compliant with OWASP Top 10 for LLM and domestic and international AI governance guidelines.