EU AI Act: European Parliament adopts the first comprehensive AI law
In one sentence On March 13, 2024 the European Parliament approves the AI Act (523 in favour, 46 against, 49 abstentions), the first comprehensive law on artificial intelligence, with a risk-based approach and specific obligations for general-purpose models. Lawyer-linguist review and Council endorsement are still pending.
The European Parliament approves the world's first comprehensive AI law. It doesn't concern a single product but sets rules for AI systems used in Europe.
The core idea is to classify systems by risk level: prohibited practices (for example social scoring, emotion recognition at work and in schools, untargeted scraping of faces from the internet), high risk (for example AI for hiring, education or essential public services), transparency duties (chatbots and deepfakes must be identifiable) and minimal risk.
General-purpose models like GPT-4 have extra documentation obligations and, for the most powerful ones, evaluation, testing and incident reporting. Fines reach up to 7% of global turnover.
Companies
European Parliament
Tools
—
Tags
Sources