NeuralStackThe European Commission released proposed amendments to the AI Liability Directive on July 21 2026. The draft establishes strict liability for high-risk AI systems and requires operators to maintain minimum insurance coverage of 50 million euros. The proposal follows six months of consultation with industry and civil society groups.
High-risk categories now explicitly include autonomous vehicles medical diagnostic tools and credit scoring systems. Companies must provide technical documentation within 30 days of an incident or face fines up to 6 percent of global revenue. The text will undergo parliamentary review starting September 2026.
The original AI Liability Directive was adopted in 2024 alongside the AI Act. Several member states including Germany and France had requested clearer liability rules for generative AI outputs. The new proposal aligns with the risk tiers defined in the AI Act that took effect in August 2025.
Why this matters
The update creates a clearer legal framework that could reduce litigation uncertainty for AI developers operating in Europe. Insurance markets are expected to launch specialized products within nine months of final adoption. Non-EU companies may need to restructure deployment strategies to limit exposure.
Final approval is anticipated by March 2027 with transposition deadlines set for December 2027.