Oxera AI policy map – October 2026

For this fifth edition of the AI policy map,¹ we have updated our database that tracks key national and supranational AI policy developments across the European Economic Area (EEA) and the UK. We have also introduced a new interface to make the map clearer and easier to use.

As with the previous editions, this curated collection brings together legal texts, strategy documents and other influential publications relevant to the governance of AI, based on the updated OECD AI Policy Observatory.²

Policies are categorised into the following four groups.

• ‘Law and legislation’: legally binding texts that were enacted by a legislative authority
• ‘Regulation’: legally binding texts from a government authority that did not require parliamentary legislative approval
• ‘Guidance on legal implementation’: instructions on how to implement a law, regulation, policy or other rule
• ‘Principles for trustworthy AI’: guidelines, propositions, codes of practice or other frameworks with voluntary participation

After substantial changes to previous editions in terms of the types and numbers of policies introduced in the EEA and UK, the collection of AI policies recorded in the OECD database now remains relatively stable. This edition includes the following four new policies.

• Two in the UK: one in the ‘Law and legislation’ category and one in the ‘Principles for trustworthy AI’ category.
• One in Greece and one in Italy: both in the ‘Law and legislation’ category.

The bar charts below the map show the number of policies held by each EEA jurisdiction and the UK, broken down by policy type. ‘Guidance on legal implementation’ remains the largest category in our curated data set—the UK has ten policies in the category and the EU has 12 at the supranational level. By contrast, ‘Law and legislation’ has the lowest country-level presence: beyond the EU itself, few individual countries have enacted their own AI-specific legislation. This reflects the EU AI Act’s role as an EU Regulation intended to support a harmonised framework across the single market and prevent divergent national AI laws. In the EEA, country-level variation is therefore more likely to arise through regulatory guidance and implementation, rather than standalone national AI legislation, except where the Act permits it or requires supporting national measures. By contrast, legislative debate over standalone AI rules is more likely to continue in non-EU jurisdictions, including the UK.