Summary
As of February 2026, the EU AI Act's provisions for general-purpose AI models are now fully in force, with high-risk financial sector AI systems facing compliance deadlines in August. Legal technology providers across Europe are racing to meet transparency and documentation requirements.
The European Union's landmark AI Act has entered its most significant enforcement phase, bringing immediate compliance obligations for legal technology providers across the continent.
As of February 2, 2025, prohibited AI practices and AI literacy obligations entered into application, marking the first major enforcement milestone. The AI Office became fully operational on August 2, 2025, taking a central role in the implementation and enforcement of the Act, particularly for general-purpose AI (GPAI) models.
For legal AI providers, the implications are substantial. The Act's risk-based classification system places many legal AI applications in the high-risk category, particularly those used in:
- Access to essential services (legal assistance)
- Administration of justice and democratic processes
- Employment and worker management
Key compliance requirements include:
1. **Transparency Obligations**: GPAI model providers must disclose training data sources, including copyrighted materials
2. **Technical Documentation**: Comprehensive documentation for models with systemic risk
3. **Human Oversight**: Mandatory human review for high-stakes legal decisions
4. **Quality Management**: Robust systems for monitoring AI outputs
The Commission's July 2025 publication of the GPAI Code of Practice provides voluntary but increasingly referenced standards for compliance. This code offers practical guidance on transparency, copyright, and safety requirements.
Penalties for non-compliance are severe: up to €35 million or 7% of worldwide turnover for prohibited practices, and up to €15 million or 3% for other infringements. These penalties apply to both EU and non-EU companies offering AI systems in the European market.
Legal industry analysts report that early adopters who began compliance preparation in 2024 are now well-positioned, while firms that delayed face significant catch-up work. The consensus is that 3-6 months of dedicated preparation is necessary for most legal AI implementations to achieve compliance.