查找相似内容

🇪🇺EU Launches AI Act Service Desk to Support Compliance The European Commission has launched the AI Act Service Desk, a dedicated platform designed to help stakeholders comply with the AI Act. The Service Desk includes an information hub providing guidance on how to apply specific provisions of the regulation. A key component of the Service Desk is the Single Information Platform, which offers interactive online tools enabling organizations to assess their legal obligations and identify practical steps to ensure compliance with the new EU AI framework. This initiative aims to give developers, deployers, and other affected entities structured support as the AI Act begins to take effect across the EU regulatory landscape. #AIAct#AIRegulation#EULaw

📖AI Agents Under EU Law: Compliance Architecture Proposal Published A new paper titled “AI Agents Under EU Law: A Compliance Architecture for AI Providers” analyzes how AI agents are regulated under the EU legal framework. The authors define AI agents as systems capable of autonomous planning, tool use, and multi-step execution with reduced human involvement, deployed across domains such as customer service, recruitment, clinical decision support, and critical infrastructure management. The paper maps regulatory obligations under the EU AI Act alongside GDPR, Cyber Resilience Act, Digital Services Act, Data Act, Data Governance Act, NIS2 Directive, Product Liability Directive, and other sectoral rules. It also integrates draft harmonised standards under CEN/CENELEC JTC 21, the GPAI Code of Practice (July 2025), the CRA standards programme (April 2025), and Digital Omnibus proposals (November 2025). A taxonomy of nine deployment categories is proposed, linking agent actions to regulatory triggers. Key compliance issues identified include cybersecurity risks (including privilege minimization outside the model), human oversight limitations due to reinforcement learning-based evasion, transparency challenges in multi-party action chains, and runtime behavioral drift under Article 3(23). The authors propose a twelve-step compliance architecture and a regulatory trigger mapping system, concluding that agentic systems with untraceable behavioral drift cannot currently meet essential AI Act requirements, and that providers must focus on exhaustive mapping of actions, data flows, systems, and affected individuals rather than classification alone. #AIRegulation#EULaw#AIAgents#AIAct#Compliance

🇪🇺📖Study Finds Limited Availability of AI Training Data Disclosures Under EU AI Act Researchers from Trinity College Dublin report that information about AI training data required under the AI Act is often missing and difficult to locate. The law requires developers to publish summaries explaining how their models were trained, using a disclosure template designed to help copyright holders enforce their rights regarding the use of copyrighted material in AI training. A pre-print study funded by Mozilla found that only a small number of such summaries could be identified. The researchers also found structural issues in accessing the disclosures. The AI Act does not specify where companies must publish the summaries, leaving the decision to developers. As a result, no common publication mechanism exists and practices vary widely. The template created by the European Commission AI Office has led to heterogeneous implementations, making it difficult to determine whether the available documents meet EU transparency requirements. Most of the identified disclosures were produced by smaller organizations, including documentation for Switzerland’s Apertus national model. A document published by Microsoft for one of its open-source models was also reviewed, but the study found that it lacked several required details. Researchers recommend creating a centralized portal for publishing transparency summaries to improve accessibility and support enforcement once the AI Act obligations become applicable in August. #AIAct#AITransparency#TrainingData#Copyright#AIGovernance#AIRegulation#EULaw

🇪🇺European Commission Releases Second Draft of AI Content Labelling Code The European Commission has published the second draft of a voluntary Code of Practice intended to help providers and deployers comply with transparency obligations under Article 50 of the AI Act. The article requires marking and labelling of AI-generated content. The updated draft reflects feedback collected in January 2026 from hundreds of stakeholders across industry, academia, and civil society, as well as input from EU Member States and representatives of the European Parliament. The revised code is designed to reduce compliance burden while promoting open standards and the use of a common EU icon for AI-generated content. It is structured in two sections: the first addresses marking and detection obligations for generative AI system providers, introducing greater flexibility and clearer guidance; the second focuses on deployers, covering labelling of deepfakes and AI-generated text related to matters of public interest with a more practice-oriented approach. Public feedback on the draft is open until 30 March 2026. The final version of the code is expected by early June 2026, while the transparency obligations under Article 50 of the AI Act will become applicable on 2 August 2026. #AIAct#AIRegulation#AIGovernance#Transparency#Deepfakes#ContentLabelling#EUlaw