Center for AI and Digital Policy Europe’s Post

🌟 #EuropeanAIOffice's First Webinar on the AI Act #AIA 🌟 30/05 at 14:00 CEST (date confirmed) Please register here https://lnkd.in/eZqxTnyY The link to the official information on the webinar is at the European AI Office web page https://lnkd.in/e_ys2MMb 🔍 #Riskmanagement logic of the #AIAct and related standards: Must-Know Fundamentals for #Regulatory #Compliance 📅 Join me at the webinar on 30 May 14:00 to 15:00 to learn: 📘 The AI Act’s regulatory logic with insights into risk and quality management systems, compliance requirements for high-risk AI systems; #RiskManagement #Compliance 🚀 Essential elements for developing risk assessments on general-purpose AI models with systemic risks; #AIRegulation #SystemicRisk 🌐 How to be ahead of the curve and future-proof your regulatory compliance with the AI Act; #FutureProofing #RegulatoryCompliance ⚖️ Whether available international standards, such as recent ISO 42001, are sufficient for the regulatory compliance under the AI Act. hashtag #ISO42001 #InternationalStandards #CEN_CENELEC 👥 Looking forward to seeing you there! #Webinar #AI European Commission Dr. Tatjana Evas Antoine-Alexandre André Josep Soler Garrido Delia Fano Yela CEN and CENELEC

A bit on short notice for many (including me): The webinar on the AI Act's risk management approach, which was delayed, is now scheduled for 2pm (CET) May 30th i.e. TODAY!! (No mention of a recording on the website

Update: This event has now been cancelled and looks like it will be rescheduled over the coming week(s). The AI Office is hosting its first webinar on the AI Act on 16 May 2024. The topic looks interesting (Risk Management Approach in the AI Act: Must know fundamentals for regulatory compliance). Sharing for info as I would have missed it without Dr. Tatjana Evas helpful post. Link to the event page: https://lnkd.in/gc5uJUY3 Link to AI Office website for those that that have been struggling to find it (like I have!): https://lnkd.in/gqveBcBz #AIAct #AIA #AIGovernance

Should be interesting to hear some first hand details about the risk management aspect of the AI Act. The webinar has been postponed though. The new date is not announced on the website just yet, so make sure you check back on the website in the next days.

[UPDATE 28/03 --> the new date is #confirmed 30/03 at 14:00 CEST] Please register here https://lnkd.in/eZqxTnyY 🌟 #EuropeanAIOffice's First Webinar on the AI Act #AIA 🌟 The link to the official information on the European AI Office web page https://lnkd.in/gBwHiD7E] 🔍 #Riskmanagement logic of the #AIAct and related standards: Must-Know Fundamentals for #Regulatory #Compliance 📅 Join me at the webinar on 30 May 14:00 to 15:00 to learn: 📘 The AI Act’s regulatory logic with insights into risk and quality management systems, compliance requirements for high-risk AI systems; #RiskManagement #Compliance 🚀 Essential elements for developing risk assessments on general-purpose AI models with systemic risks; #AIRegulation #SystemicRisk 🌐 How to be ahead of the curve and future-proof your regulatory compliance with the AI Act; #FutureProofing #RegulatoryCompliance ⚖️ Whether available international standards, such as recent ISO 42001, are sufficient for the regulatory compliance under the AI Act. #ISO42001 #InternationalStandards #CEN/CENELEC 👥 Looking forward to seeing you there! #Webinar #AI Please register here https://lnkd.in/eZqxTnyY European Commission Dr. Tatjana Evas Antoine-Alexandre André Josep Soler Garrido Delia Fano Yela CEN and CENELEC

AI System (à haut risque/high-risk) & Evaluation de conformité/Conformity assessment: contrôle interne/self-assessment or tiers/organisme notifé/third party evaluation (notified body)? 📋 The EU AI Act ((EU) 2024/1689) is part of the EU’s “New Legislative Framework” for regulating product safety. As a result, providers of high-risk AI systems must conduct a conformity assessment before placing the AI system on the market or putting it into service. 🔍The AI Act outlines two assessment methods: first, a “self-assessment” based on internal control (Article 43(1)(a) AIA), and second, an evaluation involving a notified body (Article 43(1)(b) AIA). 📊 The chart below offers a preliminary view of when each of these procedures is applicable. 💡 Note: in our third issue of the “EU AI Act in Practice” newsletter, we explore the conformity assessment process for high-risk AI systems in greater depth. #ArtificialIntelligence #AIAct #AIRegulation #AICompliance #EYLaw

The EU AI Act lays out procedures for assessing the compliance of AI systems with the regulation's requirements. Though much gray still exists, some key points regarding auditing adherence include: 1. High-risk AI systems must undergo a conformity assessment procedure carried out by the provider either through self-assessment or with the involvement of a third-party notified body, depending on the system. This involves examining the system's technical documentation and quality management system to verify compliance. 2. Providers must keep detailed technical documentation on their high-risk systems to demonstrate compliance. This documentation can be accessed by national authorities upon request. 3. Market surveillance authorities in Member States are empowered to verify compliance of systems, request documentation/information, carry out audits, and take corrective actions if needed. 4. There is a post-market monitoring system requirement for high-risk systems - providers must actively collect data on the performance of deployed systems to evaluate continuous compliance. 5. There is an EU database for registration of high-risk systems that contains information to identify them and assess compliance. Market surveillance authorities can access this to support oversight. 6. General-purpose AI models also have transparency requirements and obligations on technical documentation. Their compliance is monitored by the EU AI Office which can request documentation, carry out evaluations, and request measures from providers if needed. 7. There are administrative fines and penalties laid out in the regulation for non-compliance, aimed to incentivize compliance from providers/deployers. To summarize - a combination of conformity assessments, technical documentation requirements, post-market oversight powers of authorities, transparency obligations, and penalties/fines makes up the compliance and auditing system. 🌐 For more info: https://lnkd.in/d4bYGBT6 #euaiact #aicompliance #ethicsinai #align #TheBusinessofCompliance #ComplianceAlignedtoYou

Congratulations FAIRLY AI on this important work and for being leaders in Canada in the development of proper governance around AI in machine learning models. For boards directors, this is your new cyber security question …. Now it’s time to ask, how machine learning models are protected against bias and what governance is built into your company’s AI strategy? Set high ethical standards.

We are honoured to have been selected to work with ATB Financial, Standards Council of Canada (SCC) and the Responsible AI Institute in the first pilot for ISO/IEC 42001 in Canada. By supporting industry standards like ISO/IEC's and RAII's conformity assessments, our AI-native compliance solution enable organizations to adopt predictive and generative AI with trust, risk and security management (AI TRiSM) under control. Visit www.fairly.ai today to find out why industry experts and analysts such as IDC, Gartner and S&P Global 451 Research consistently named FAIRLY AI as a representative vendor and major player in the AI TRiSM, AI Governance, Risk and Compliance space. #aitrism #AIcompliance #AIGRC

Why should you pay attention to new #AIStandards? They provide frameworks for effective AI governance and tracking them can provide advance notice of potential future legal requirements. Yesterday, ISO/IEC 42001 was released with guidance on developing, improving, and monitoring Artificial Intelligence Management Systems (AIMS). This standard will likely guide emerging regulations – primarily in the European Union and United States – in the form of guiding principles and mechanisms for future legislation. Read more from our expert AI policy team below 👇 https://lnkd.in/ePTxa2EE