Catalogue of Tools & Metrics for Trustworthy AI

These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.

Behavior Elicitation Tool



Behavior Elicitation Tool

Behavior Elicitation Tool (BET) is a complex-AI system that systematically probes and elicits specific behaviors from cutting-edge LLMs. Whether for red-teaming or targeted behavioral analysis, this automated solution is Dynamic Optimized and Adversarial (DAO) and can be configured to test the robustness precisely and help to have a better control of the AI system. The system's capabilities are continuously enhanced by research in AI interpretability and safety.

This automated solution enables precise specification and elicitation of any target behavior, from security vulnerabilities to desired output patterns. Leveraging advanced prompt engineering and behavioral steering techniques, BET systematically maps the complete behavioral landscape of LLMs and their deployment contexts (system prompts, scaffolding, RAG systems, output filters, etc.).

BET's architecture enables granular control over model behavior throughout the entire AI development lifecycle. By integrating into CI/CD pipelines during training, fine-tuning, and deployment phases, it provides dynamic behavioral verification across arbitrary target specifications. Unlike static benchmarks, BET exploits model-specific characteristics to identify and shape behaviors across the full spectrum of possible outputs, enabling developers to precisely tune their models' responses for any desired application context.

The system's sophisticated behavior mapping capabilities are continuously enhanced by breakthroughs in AI interpretability and safety research, providing unprecedented control over LLM outputs. This enables more robust behavioral confidence for industrial GenAI applications while maintaining visibility into model dynamics.

Use Cases

There is no use cases for this tool yet.

Would you like to submit a use case for this tool?

If you have used this tool, we would love to know more about your experience.

Add use case
catalogue Logos

Disclaimer: The tools and metrics featured herein are solely those of the originating authors and are not vetted or endorsed by the OECD or its member countries. The Organisation cannot be held responsible for possible issues resulting from the posting of links to third parties' tools and metrics on this catalogue. More on the methodology can be found at https://oecd.ai/catalogue/faq.