About
Giskard is a French SME that develops an open-source platform for evaluating and red-teaming AI systems. Our technology addresses a critical need in the AI value chain: ensuring that AI models and agents are robust, reliable, safe, and compliant before deployment.
Our core capabilities include automated vulnerability detection in LLMs and generative AI systems (hallucinations, prompt injection, data leakage, bias), evaluation of RAG pipelines and autonomous AI agents, multi-agent system testing and robustness benchmarking, and AI Act conformity assessment tooling.
Giskard's open-source library is widely adopted by the European AI community, with thousands of users across research and industry. Our enterprise platform, Giskard Hub, provides continuous red-teaming for production AI systems.
We are seeking consortium partners for two Horizon Europe Cluster 4 topics under the AI, Data and Robotics Partnership:
-
HORIZON-CL4-2026-05-DIGITAL-EMERGING-02 — Next-Generation AI Agents: We bring evaluation and benchmarking methodologies for autonomous agent robustness, safety, and multi-agent coordination.
-
HORIZON-CL4-2026-04-DIGITAL-EMERGING-19 — GenAI4EU Booster: We provide the trustworthy AI evaluation layer for GenAI deployment in priority sectors.
We can lead or contribute to workpackages on evaluation, benchmarking, red-teaming, and regulatory compliance.
Digital technologies
Artificial Intelligence
Digital Innovation Hubs
DIH - Artificial IntelligenceDIH - Cybersecurity
European Partnerships
Artificial Intelligence, Data & RoboticsMade in Europe
Social sciences and humanities
Human-centric approachesLaw
European Data Spaces
HealthResearch and Innovation