European Data Protection Board’s Post

New tool is out: Check-list for AI Auditing to help evaluate the GDPR compliance

I’m still coming to terms with the fact that the AI Act Comprehensive track at #GPS24 attracted such a massive level of attention. Here are two key slides of my presentation trying convey (in the simplest possible way) the level of granularity attached to different types of #AI. Essentially, the AI Act establishes six different levels of risk depending on the type or purpose of the AI technology in question, namely: 1️⃣Prohibited AI 2️⃣High risk AI systems 3️⃣Transparency risk AI 4️⃣General-purpose AI models 5️⃣General-purpose AI models with systemic risk 6️⃣All other AI not covered by the above If you compare that to the non-personal data v. personal data v. special category personal data categorisation of the #GDPR, then you get a sense of how complex the regulation of AI technology is going to become.

🔔 Join the first Research Policy and Standards webinar of 2024 👉 https://lnkd.in/dEDJJMwG to find out about the ICO's #generativeai consultation as well as researching sensitive subjects and trigger warning content. 👉 Information Commissioner's Office (ICO) consultation on Generative AI and #data protection Launched on 15 January 2024, the ICO Generative AI and data protection consultation series focus on how aspects of data protection law should apply to the development and use of Generative AI models. Book your place to get updated on the developments in this area whilst also get up to speed on researching sensitive subjects and trigger warning content. 👉MRS Standards and Policy team Our quarterly standards webinars, designed to help you stay current and work within the rules, are compiled and presented by the MRS Standards and Policy team, the experts behind Codeline, the Code of Conduct, GDPR, regulations and guidelines. #webinar #membres #marketreseaerch #researchpolicy #researchstandards #mrx

𝗖𝗼𝗺𝗽𝗹𝗶𝗮𝗻𝗰𝗲 𝘂𝗻𝗱𝗲𝗿 𝘁𝗵𝗲 𝗘𝗨 𝗔𝗜 𝗔𝗰𝘁: 𝗦𝘂𝗺𝗺𝗮𝗿𝘆 𝗮𝗻𝗱 𝗞𝗲𝘆 𝗜𝘀𝘀𝘂𝗲𝘀 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞𝐬 𝐧𝐨𝐭 𝐨𝐧𝐥𝐲 𝐟𝐨𝐫 𝐃𝐚𝐭𝐚 𝐄𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐢𝐧𝐠 𝐚𝐧𝐝 𝐃𝐚𝐭𝐚 𝐒𝐜𝐢𝐞𝐧𝐜𝐞 The EU AI Act introduces strict regulations that categorize AI based on risk levels and require compliance and transparency from developers and users alike. Challenges include the cost of compliance inhibiting innovation, the different approaches to compliance assessment, the interaction risks between multiple AI systems, the lack of guidance for human oversight, the interpretability of transparency obligations and the interplay with the General Data Protection Regulation. Data engineers must ensure careful data management, security and governance to meet the new requirements. Data scientists face the challenge of developing ethical AI models that meet transparency standards and mitigate bias. 𝗥𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝘀: 𝗥𝗼𝗯𝘂𝘀𝘁 𝗖𝗼𝗺𝗽𝗹𝗶𝗮𝗻𝗰𝗲 𝗦𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗲𝘀: Implementing comprehensive documentation and cybersecurity measures to meet regulatory standards. 𝗘𝘁𝗵𝗶𝗰𝗮𝗹 𝗔𝗜 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁: Mitigating biases in AI models and ensuring transparency in decision-making processes to comply with transparency obligations. 𝗘𝗻𝗵𝗮𝗻𝗰𝗲𝗱 𝗥𝗶𝘀𝗸 𝗠𝗮𝗻𝗮𝗴𝗲𝗺𝗲𝗻𝘁: Conducting thorough risk assessments and integrating human oversight mechanisms to mitigate potential risks effectively. Addressing these challenges requires clear guidelines, harmonized standards and continuous stakeholder engagement. Following these recommendations will not only ensure compliance with EU AI Act, but will also promote trust in AI technologies and their reliability. #AIregulation #EUAIAct #ArtificialIntelligence #DataEngineering #DataScience #GDPR #EthicalAI #Innovation #Compliance #EUregulations #GenerativeAI https://lnkd.in/d-B77swz

With the new #EUAIAct coming into action and being on my mind, my few cents on it: This Act is set to revolutionize data privacy in the AI landscape. This brings stricter regulations around high-risk AI systems. And would the same data privacy requirements apply? if yes, is it feasible? Well, their will be no relaxation of data protection principles of GDPR like transparency, limiting practices and safeguarding of individual's personal data and fundamental rights. In fact, it would become even more important to comply with data privacy and security practices especially when it comes to high-risk AI systems. Rules will be less stricter for low-risk AI systems. This actually marks a significant step towards a more ethical and privacy-centric approach to AI development in the EU and beyond. #AI #DataPrivacy #EURegulation #AIethics

And it is a fact! The EU AI Act has finally been approved by the Council of Ministers. The risk-based approach has largely won out in the world's first AI regulation. And that is a good thing™. I believe that regulation should always serve to mitigate clearly understood and well-defined risks. And this is what we got. Regulation should also focus on reacting to violations -- i.e., legal actions against perpetrators -- rather than red tape and bureaucracy in the implementation, i.e., forcing businesses and individuals to prove a priori that they will not break the law. How the AI Act is implemented will determine that. The GDPR implementation serves as a good example of how *not* to do it. Anyway, the law is here now, and hopefully the *risks* associated with AI is significantly diminished by it.

Getting ready for a free webinar next week 21 March at 11.00 on AI as a DPO for RESPONSUM. To be clear, not another "What is the AI Act" session but a pragmatic approach to handling AI challenges as a DPO. Together with 🦉Bavo Van den Heuvel, he and I reflect on how to approach AI tools as you are confronted with them as a DPO. How we can review, monitor or challenge AI data protection risks as a DPO without being an AI Expert yourself. We take it back to the basic principles, risks, and practical examples how we have addressed these data protection issues in an AI context when we encountered them. Register here: https://lnkd.in/ee27-ufz

📌 Andersen's IP, IT, and Data Protection Service Line is excited to announce the release of their new publication, EU Artificial Intelligence Act: Key Considerations for Data Protection Officers. Explore the guide 👉 https://bit.ly/4bIedcDhttps://lnkd.in/dpAStGAG> This insight explores the implications of the groundbreaking EU #AI Act—the first regulation addressing AI technology's untapped potential. Learn more about how the Act ensures personal data protection through existing data laws and introduces new obligations aimed at safeguarding data in high-risk AI systems. It also highlights the vital role of Data Protection Officers (#DPOs) in ensuring compliance, transparency, and risk management, in alignment with GDPR principles. https://lnkd.in/d4X2DceV #AI; #DPO; #Andersenprivacy; #GDPRAI

EU AI Act - Quick Factsheet The World's first comprehensive framework for regulating the uses of artificial intelligence has been agreed by the EU. The final text is yet to be finalised so what do we know so far? 📌 It's likely the Act's extraterritorial scope will extend to any AI systems placed on the market or put into service in the EU. 📌 Specific AI uses which threaten democracy and people's rights will be banned, with certain exemptions in the areas of law enforcement and national security. 📌 The requirements organisations need to meet will be tiered dependent on the risk AI systems pose. Stringent measures will be required for AI systems classed as high risk. 📌 Non-compliance could lead to fines much higher than under GDPR ☛ Read more in our quick EU AI Act Factsheet: https://lnkd.in/eBJru9jn #aiact #artificialintelligence