Mission
Promote, develop and coordinate interdisciplinary research in Trustworthy and Reflective AI: reliable, explainable, auditable, safe systems aligned with human agency, responsibility, fundamental rights and public interest.
Governance
Non-profit by statute. Independent by design. Built to collaborate.
TRAILAB's non-profit structure supports long-horizon research, cross-institutional collaboration and reinvestment in scientific capacity. Its governance is designed to protect scientific independence: partners and funders support programmes, but they do not determine research conclusions.
Statutory foundation
Statutory foundation
Activities
Basic and applied research, methodologies, validation, evaluation, reports, policy work, training, consortia, contracts, agreements, services and knowledge transfer linked to the association's purposes.
Independence
Scientific, ethical and political independence is protected. Funding, sponsorship, collaboration or contracts must not condition research results, technical conclusions or ethical positions.
Mission-driven execution
Resources may come from research funds, collaboration agreements, services, training, reports and sponsorship. Applied work supports the mission by creating evidence, methods and reusable capacity.
Category clarity
Members, collaborators, funders, clients, advisors and project partners are not the same category and do not automatically receive governance rights.
Transparency
Project-level accounting, ethics, conflicts, confidentiality and publication rules are part of the governance infrastructure.
Legal and standards translation
Legal and standards translation
EU AI Act
Risk classification, prohibited practices, AI literacy, GPAI governance and high-risk obligations are treated as design constraints that must become evidence, documentation and oversight.
Harmonised standards
CEN-CENELEC and ETSI standards translate legal requirements into technical specifications for risk management, data governance, logging, transparency, human oversight, robustness, cybersecurity, quality management and conformity assessment.
ISO/IEC 42001
AI management systems provide an organisational structure for responsible development, provision and use of AI systems.
ISO/IEC 23894
AI-specific risk management supports identification, analysis, evaluation, treatment and monitoring of AI risks across the lifecycle.
Sector frameworks
Automotive, aviation, industrial automation, medical and other regulated domains require domain-specific assurance, not generic AI governance.
Fundamental rights
Legal compliance is incomplete unless impacts on rights, contestability, remedy and human oversight are mapped to real decisions.
Scientific independence
Research conclusions and public evidence must not be controlled by funders, clients or project partners.
Applied public value
Projects should create reusable scientific, technical, legal, educational or institutional capacity beyond one-off delivery.
Category clarity
Members, funders, clients, collaborators, project partners and advisors must not be blurred.
Human agency
Governance is not only compliance. It must preserve responsibility, contestability, creativity and explicit human commitment.
Transparency by project
What can be public should be documented; what is confidential must be protected without turning claims into marketing.
Mission lock
All economic activity is instrumental to the non-profit mission and cannot privatise the lab's purpose.