Project Babel · AI Governance Observatory

jurisdiction readout

18 analyzed sources · 2024-07-12 to 2026-03-20

European Union's current AI governance record is guidance-heavy, with European Commission and European Commission / AI Office appearing most often in the source base. Across the analyzed documents, Fundamental rights is the clearest rationale, with Consumer and public safety as the next strongest layer, while transparency and disclosure, model documentation, and risk assessment recur most often in the operative expectations.

governance posture

The source base is guidance-heavy, so implementation detail appears more often than hard-edged statutory obligations. The strongest institutional signals come from European Commission, European Commission / AI Office, and AI Office, and the corpus is weighted toward regulatory guidance and standards guidance.

implementation

Operationally, the sources most often point to transparency and disclosure, model documentation, and risk assessment. When the corpus gets concrete about consequences, it most often references regulatory supervision and reporting obligations.

source coverage

This readout is based on 18 analyzed documents spanning 2024-07-12 to 2026-03-20. The corpus is weighted toward regulatory guidance and standards guidance. On the binding side it leans toward regulatory guidance and voluntary commitment, so it captures official policy posture more directly than downstream enforcement practice. The most recent additions in the current mix are Standardisation of the AI Act and AI Pact.

frame distribution

99% of framings observed

33385165272

XRSSECRGTECOINVSOCSAFLAB

framing landscape

Fundamental rights

Talks about AI as a risk to people's rights.

Used when a document explains its rules through protection of individual rights — non-discrimination, privacy, dignity, due process, or the right to a remedy.

Fundamental rights is the clearest rationale in the corpus, with Consumer and public safety and Innovation enablement still materially shaping how governance is justified.

operational profile

recurring requirements and consequences

top safeguard requirements

Transparency and disclosure16 · 89%
Model documentation14 · 78%
Risk assessment12 · 67%
Human oversight10 · 56%

top enforcement hooks

Regulatory supervision16 · 89%
Reporting obligations14 · 78%
Soft enforcement11 · 61%
Civil penalties5 · 28%

key sources in this readout

selected from the analyzed corpus

Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence2024-07-12

LegislationBinding lawFundamental rights

Statutory anchor for risk assessment and transparency and disclosure, with regulatory supervision as the clearest consequence or oversight hook.

Questions & Answers on the AI Act2026-01-28

Regulatory guidanceRegulatory guidanceFundamental rights

Regulatory implementation anchor for risk assessment and pre-deployment testing, with civil penalties as the clearest consequence or oversight hook.

Template for general-purpose AI model providers to summarise their training content2025-07-24

Regulatory guidanceBinding lawFundamental rights

Statutory anchor for transparency and disclosure and model documentation, with civil penalties as the clearest consequence or oversight hook.

Commission publishes the Guidelines on prohibited artificial intelligence (AI) practices, as defined by the AI Act.2025-02-04

Regulatory guidanceRegulatory guidanceFundamental rights

Regulatory implementation anchor for risk assessment and human oversight, with regulatory supervision as the clearest consequence or oversight hook.