Open-Source AI Testing and LLM Vulnerability Scanning

Giskard provides advanced security and trustworthiness solutions for generative AI systems, focusing on automated vulnerability detection and proactive risk management. Its platform identifies hallucinations, unsafe model behaviors, data leakage risks, and ethical concerns by analyzing both internal enterprise data and external cybersecurity intelligence.

The company’s system automates test generation using proprietary algorithms to evaluate models against client-specific datasets, simulating edge-case failures and adversarial behaviors. Giskard also conducts continuous threat modeling through red team-style adversarial evaluations, targeting vulnerabilities such as prompt injection, bias amplification,

stereotyping, discriminatory outputs, and private data exposure. Additional cross-domain risk assessments help organizations detect harmful content generation and privacy violations.

The platform includes collaborative tools that allow engineers, data scientists, and compliance teams to contribute directly to AI Security audits. Giskard integrates security validation into CI/CD pipelines, combining automated scanning with human-in-the-loop analysis to prioritize vulnerabilities by business impact. Through this approach, Giskard supports evolving global AI safety standards and contributes to the broader technical community through partnerships and open evaluation frameworks.

Giskard secured a €1.5 million round of funding, led by Elaia with participation from Bessemer Venture Partners and unnamed angels, in December 2025.