European experts convene in Stavanger to harmonise AI standards

The CEN-CENELEC Joint Technical Committee 21 (JTC 21) concluded a plenary meeting in Stavanger, Norway, from May 12–15, 2025, marking a significant advance in European artificial intelligence (AI) standardisation. Hosted by Standards Norway at the Havtil venue, the event brought together more than 300 experts from 24 countries, alongside over 1,000 specialists from national mirror committees, to accelerate the development of harmonised AI standards crucial for the EU AI Act and Europe’s digital future.

The Stavanger plenary underscored both the urgency and complexity of harmonising AI standards across Europe. These standards, mandated by the European Commission, are designed to provide organisations with a clear path to compliance with the EU AI Act, especially as the Act’s high-risk AI provisions are set to take effect in August 2026. Legal certainty and robust frameworks are vital for businesses and public authorities navigating the evolving regulatory landscape.

JTC 21’s collaborative approach, involving stakeholders from industry, academia, government, and civil society, ensures that the standards reflect a broad consensus and address diverse needs. This inclusiveness is seen as essential for the legitimacy and wide acceptance of the resulting standards.

Key outcomes from Stavanger

Advancement of draft standards:
Significant progress was made on draft European Norms (ENs) supporting the AI Act. These drafts will undergo further refinement based on national feedback throughout 2025, with finalisation targeted for 2026.

Inclusiveness and civil society rngagement:
The plenary reaffirmed the importance of involving civil society, making sure AI standards uphold equality, human rights, and the ethical principles required by the EU AI Act.

International collaboration:
JTC 21 strengthened its cooperation with global standardisation bodies such as ISO and IEC, ensuring European AI standards are internationally relevant and support EU trade and innovation objectives.

Road ahead

While the original timeline aimed for completion of key standards by August 2025, the current project plan anticipates that the work will extend into 2026. This extension allows for thorough consensus-building and quality assurance, but also means companies will have less time to implement the standards before the AI Act’s high-risk provisions come into force in August 2026.

The committee is taking extraordinary measures to streamline development, working closely with the European Commission’s AI Office to align on priorities and ensure that the standards reflect the state of the art and the needs of European stakeholders.