Dydon AI’s Post

☑ CHECKLISTS FOR COMPLIANCE WITH THE AI ACT With the #AIAct of the #EU taking effect today, companies are well advised to start preparing and ensure #compliance with the new requirements (see our latest LinkedIn post on the transition periods in the comments). To help you with this, we have put together a #checklist outlining the obligations for various operator roles, which you can find on our blog: https://lnkd.in/gvk3ympn As your obligations depend on the qualification of your AI tools as an #AISystem and your respective operator #role, we recommend a prior case-by-case assessment. Our #SelfAssessmentTool can help you with this process.  Cornelia Stengel Gaspare T. Loderer Jonas Tresch Vera Vallone Kellerhals Carrard

MLL Legal summarises the latest news on #airegulation in the #europeanunion. In our latest blog post, we look at what the trilogue's political agreement on the EU's #aiact means for the upcoming final version of the law, which will need to be carefully scrutinised. To provide some insights #frompolicytopractice, we highlight the importance of the work of CEN and CENELEC Joint Technical Committee 21 #artificialintelligence, which builds on the existing work of ISO/IEC Artificial Intelligence (AI) JTC1 SC42. We explain why Swiss companies using artificial intelligence systems should follow it closely and ramp up their #aicompliance and what the latest developments on regulating #ai are in #Switzerland. The co-authors are Lukas Bühlmann and Michael Reinle. Thanks also to Michael Schüepp for his support. https://lnkd.in/dtvVPBer

🌟 Exciting News Alert! The AI-Act has been officially published today! 🚀 📆 First Important Deadline: Starting on 2 August 2024 (20 days after publication), businesses need to gear up for the first key deadline: The AI literacy rules will apply 6 months after the AI-Act comes into force, i.e. 2 February 2025. This mandates providers and deployer (= user in the course of professional activity) of AI systems to ensure that their staff and others involved in the operation and use of AI systems on their behalf have sufficient knowledge and training related to the use of AI systems 🔍 Our Recommendation: It's crucial to establish tailored policies and guidelines within your organization to align with the new regulations. Stay informed and check out the official publication here: [Link to the AI-Act publication](https://lnkd.in/dJyGyCQS) #AI #ArtificialIntelligence #Regulations #Compliance #Business #Training #Policies #Guidelines

Just published: EuZ 03/2024 with a contribution on the EU's latest digital regulation (#DMA, #DSA, #DA, #DGA, #EHDS, #AI Act) by Chayanis Aueamnuay, Carmen B., Stella G, Luca Graf and Andreas Heinemann: The EU is responding to the challenges of the digital transformation with an unprecedented number of regulatory acts. The central objectives of the Union, such as the creation of the single market, and its values shall also apply to the digital world. By adopting landmark regulations, the Union wants to strengthen its digital sovereignty instead of following the rules of others. #eizpublishing #eizuzh #europainstitut #openaccess #ejournal

📰 Happy to share that I’ve been featured in NZZ Neue Zürcher Zeitung, with my comments on the new SB-1047 Safe and Secure Innovation for Frontier Artificial Intelligence Models Act introduced in California. The bill already caused a LOT of debates with Elon Musk taking a pro-regulation approach, while Nancy Pelosi is warning about over-regulation. (Yes, you read that right.) And, well, I just can't sit still for 5 minutes, can I? My take: 👉 Concerns seem a bit overblown as the law narrowly defines 'covered models' (AKA trained using a LOT of computing power (10^26 FLOPS) or the cost of which exceeds $100,000,000) and the developers of such 'covered models' also have the resources to comply with the obligations like keeping unredacted safety and security protocols or conducting annual audits. 👉 The Bill would also prohibit the use of models with the risk of causing critical harm. Such harms include models enabling the creation of nuclear, chemical or biological weapons or the use of which can cause mass casualties or at least $500,000,000 of damage. I think we can all agree that if a 'covered model' enables similar harms it should probably be taken down. 👉 The bill authorizes the State Attorney General to bring a civil action and protects whistleblowers by demanding reporting mechanisms and prohibiting retaliation. (Some actual enforcement possibilities, refreshing.) 👉 It demands establishing a Board of Frontier Models and a CalCompute consortium to foster ethical AI innovation. (You can never have too many heads looking at the same problems, or?) Full text of the Bill available here: https://lnkd.in/dsAQ4ZJV Article (in German) available here: https://lnkd.in/d5tFzB4v

🌐 Embracing the transformative power of #ArtificialIntelligence is central for Switzerland's 🇨🇭 economic prosperity. As we navigate the forefront of technological evolution, it is crucial for #Switzerland to fully integrate AI's capabilities into its (economic) strategies. Neglecting this imperative may lead to relinquishing our leadership in innovation and diminishing our leading position in global economy. 🔍 In the sphere of #AI #regulation, economiesuisse emphasizes the importance of a nuanced approach. Advocating for principle-based and technology-neutral regulation, economiesuisse recommends precise adjustments to existing frameworks, ensuring they remain agile and future-ready. This strategic approach will cultivate an environment conducive to AI's #growth, all the while maintaining the highest ethical standards. 💬 Engage in the dialogue with economiesuisse on the basis of its position paper #dossierpolitik by Erich Herzog & Leonie Ritscher: https://lnkd.in/edcKrWuE (German) #Innovation #Technology #SwissEconomy 🚀🌍

At the end of last week, the sixth ministerial meeting of the Trade and Technology Council (“#TTC”) took place in Leuven, Belgium. The joint statement is worth a read: https://lnkd.in/dhSy9sWd This important gathering demonstrated the wide variety of areas in which the EU and US are cooperating including Artificial Intelligence, #Quantum Computing, #6G, critical minerals, #semiconductors, green energy and more. The EU and US clearly wants to cooperate in standards development in new and critical technologies. Divergent standards will not benefit consumers or industry and it is great to see this strong efforts in additive manufacturing, #AI and more. U.S. Department of Commerce U.S. Commercial Service

Today the EU "AI-Act" was published. You can read it here: https://lnkd.in/d2r2gmRr I guess playtime is over and everyone using AI in a professional context has to play by these rules from now on (at least if you're doing business in the EU).

Last week, the European Union unanimously passed a first-of-its-kind comprehensive regulation on AI which will help promote human-centred and trustworthy AI in Europe—a major milestone for AI governance in Europe. Joanna Bryson had signed a petition together with a group of academics, scientists, think tanks, businesses and civil society calling for Germany to vote "YES" to the legislation. The petition made the point, among others, that an EU AI Act as a foundation for governing AI across the bloc is a better choice for businesses and society than no regulation at all which could also harm the bloc's competitive edge for AI innovation. It turns out the German representation agreed. Read about why this is important in this Tagesspiele article (in German) here: https://lnkd.in/dpMGa8jS #EUAIact #aigovernance

When I talk to contacts in my international data centre network, they are all amazed at the speed with which the member states of the European Union have passed the "AI Act", the world's first law for uniform rules on artificial intelligence in the EU. In principle, the law is intended to protect citizens in the EU from particularly abusive AI. Is it a good thing - definitely yes! In my humble opinion, it is by no means overzealous or overregulatory, but surprisingly differentiated. The AI Act regulates 4 categories: Firstly, there are applications that are generally prohibited, such as social credit systems, which some countries use to grade and harass their citizens. Secondly, high-risk applications, such as decisions on asylum applications or job applications, are subject to particularly strict controls. Third in the regulatory pyramid are programmes with limited risk, e.g. bots such as ChatGPT, whose providers must disclose whose data they use to train their AI. And finally, there are unregulated services such as spam filters for emails. Although the AI Act has now been in force since August, the law will be fully applicable in 2 years. The EU is therefore giving companies, especially the small, generally European providers, two years to adapt to the law. As always, I am happy to discuss this topic with you. It is therefore best to book an appointment via my profile..😉. jon LABAN Tobias Dehmel Michael Tuttas Matthias Haymoz Babak Falsafi Andrew Grace Christian Kallenbach Serdar Gorali Pascal Seliger Jon Summers Tor Björn Minde Christoph Streit Antti Laine Jeroen Burks Raúl Álvarez Ross Verrall Colm Shorten Joe MacDonald Thomas Pfuetzenreuter Dr. Béla Waldhauser Mark Acton Jonathan Evans Daniel Satzinger Deborah Andrews Isabelle Kemlin Francois Tournesac Jaime Comella Gómez-Aller David G. Dr. Annette Doms Julie Bolthouse Elena Schlossberg Henrik Børling John Booth MBCS, CDCAP, CDCSP #moderndatacenters #hpc #cloudnative #combinedheatcompute #chc #sustainability #energyefficiency #decarbonization #scope3 #renewableenergy #circulareconomy #opensourcesoftware #heatharvesting #opensourcehardware #greendeal #rzdacheu #esg #esgreporting #nachhaltigkeitsreporting #certification #cybersecurity Link: https://lnkd.in/dq3x8QRG