Articles on AI in legal practice, the EU AI Act, Nordic law and the legaltech market.
Hallucinations are the biggest risk in legal AI. Three layers of protection: every legal claim bound to a verified primary source, instructed uncertainty ("say 'I don't know' rather than fabricate"), and extended thinking for multi-step reasoning. Plus an honest look at what the layers do NOT protect against.
ZDR explained at two layers: the the AI provider enterprise contract + LexCodex's own retention policy. The difference between "we don't train on your data" and "Zero Data Retention". What ZDR does NOT protect against. Concrete questions you as a lawyer can ask any AI vendor.
"Language toggling" isn't enough — citation systems, court hierarchies and primary sources differ. How LexCodex's per-jurisdiction architecture handles Swedish + Norwegian law + the EU/EEA framework. Roadmap for Denmark, Finland, Iceland.
The EU AI Act takes full effect on 2 August 2026. For law firms and in-house teams it raises two concrete questions: is your AI tool an "AI system" under the regulation, and how do you classify your clients' AI systems? A walk-through of the risk classes prohibited, high-risk, limited risk and minimal risk — plus a concrete checklist before 2 August 2026.
