Quality Assurance for AI

Projektinformationen

Giskard

ID Finanzhilfevereinbarung: 190198093

DOI

10.3030/190198093

Projekt abgeschlossen

EK-Unterschriftsdatum 22 August 2023

Startdatum 1 September 2023

Enddatum 31 August 2025

Finanziert unter

• The European Innovation Council (EIC)

Gesamtkosten

€ 3 729 395,43

EU-Beitrag

€ 2 499 999,99

Koordiniert durch

GISKARD AI
France

Periodic Reporting for period 2 - Giskard (Quality Assurance for AI)

Berichtszeitraum: 2024-07-01 bis 2025-08-31

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

The rapid adoption of AI technologies has precipitated a corresponding rise in concerns regarding ethical biases, prediction errors, and cybersecurity risks. Current AI quality tools predominantly rely on manual testing methods, creating a gap that strains the capacity and resources of AI/ML engineers amidst escalating demands.

In response to these challenges, Giskard is developing a comprehensive solution comprising both open-source and Software-as-a-Service (SaaS) platforms. These platforms are specifically designed to enhance the quality assurance processes for AI models across industries. Giskard’s initiative focuses on automating AI Quality and Security Testing, thereby empowering organizations to effectively manage and mitigate risks associated with AI deployments. This abstract outlines Giskard’s commitment to advancing AI quality assurance practices through innovative technological solutions, aimed at addressing critical challenges in AI deployment and ensuring the reliability and ethical integrity of AI systems in practical applications.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

During the EIC project, Giskard has successfully completed its Quality and Security testing platform for AI models for enterprise and open-source community. Beyond technical milestones, Giskard has also secured 2 patents and conducted extensive communication and dissemination activities to broaden its market presence.
Key accomplishments include:
Development of open source scan features to detect vulnerabilities and biases in AI models. These scans cover LLM, NLP, tabular, and computer vision models.
Development of an enterprise platform for scanning and testing Gen AI agents developed by large companies.
Development of a compliance platform to assess the conformity of the AI development against the EU AI Act.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

The development of GISKARD's open-source and SaaS solution for AI quality assurance represents a significant advancement, addressing critical gaps in current manual testing. By automating AI Quality Testing, Inspection, and Remediation, GISKARD aims to enhance reliability and mitigate risks such as ethical biases and cybersecurity vulnerabilities.

Potential impacts include streamlining AI deployment across industries, fostering innovation, and building stakeholder trust. Key needs for further uptake and success include ongoing research and development to keep pace with technology, demonstrating effectiveness across diverse applications, accessing markets and finance for scalability, robust commercialization strategies, strong intellectual property rights protection, international expansion, and advocating for supportive regulatory frameworks.

Addressing these needs will maximize GISKARD's impact, accelerate market adoption, and establish it as a leader in ethical AI deployment worldwide.

Mitigate AI Risks with our holistic platform for AI Quality, Security and Compliance