Europe Launches KnoWare

Trondheim, Norway, 9 – 11 June 2026. The KnoWare project held its official Kick-Off Meeting in Trondheim, Norway, marking the launch of a Horizon Europe initiative to develop the first scientific methodology for detecting and governing emerging capabilities in General Purpose AI (GPAI), the powerful AI systems increasingly shaping science, industry, and society.

General Purpose AI models develop unexpected new abilities as they scale in size and complexity. These “emerging capabilities” appear seemingly without warning, surprising developers, regulators, and users alike. Today, the field can only react after a capability has already surfaced. KnoWare aims to change this fundamentally: by building the scientific tools to detect early signs of a capability forming, before it fully materialises, and by equipping organisations and regulators to act in advance.

Hosted by SINTEF, the three-day Kick-Off Meeting brought together all 13 consortium partners from 7 European countries and Canada. Partners aligned on scientific objectives, governance structures, and the project’s roadmap for developing its core innovation: a “prefiguration methodology”, a framework for identifying partial behaviours, latent patterns, and enabling conditions that signal a new AI capability is forming.

KnoWare will deliver eight Key Exploitable Results spanning the full spectrum of AI governance: a GPAI capabilities roadmap, an open-source tool suite including the RiskPath Mapper and FairScope bias detection tools, novel benchmarks for risk-aligned evaluation, socioeconomic and environmental impact frameworks, and a regulatory alignment roadmap providing direct operational guidance for compliance with the EU AI Act. These results will be co-designed with industry through four real-world use cases:(a) energy & commodity trade, (b) biological hazards & pandemic prevention, (c) maritime cybersecurity & critical infrastructures, and (d) chemical hazards in food value chain.

The project’s multidisciplinary consortium unites research excellence with legal expertise, technological innovation, and industrial application. Partners include SINTEF (coordinator, Norway), RAINNO (DEC Manager, Greece), MEWS Partners and MEWS Labs (France), NTNU (Norway), KU Leuven (Belgium), Aristotle University of Thessaloniki | AUTH (Greece), University of Granada (Spain), University of Warwick (UK), Netcompany (Denmark), CAIRNE (Ireland), BIA Analytical (UK), and the University of Guelph (Canada).

With a total budget of € 3,999,862.50 funded by the European Union under Horizon Europe Grant Agreement No. 101298802, KnoWare represents a decisive investment in keeping Europe at the forefront of trustworthy, well-governed AI. The project’s work is directly relevant to the EU AI Act, which classifies certain GPAI systems as posing systemic risk, a risk category that KnoWare’s methodology is uniquely designed to assess and anticipate.