VeraSafe’s Post

🇪🇺 𝗧𝗵𝗲 𝗘𝗨 𝗔𝗜 𝗔𝗰𝘁: 𝗔 𝗡𝗲𝘄 𝗙𝗿𝗼𝗻𝘁𝗶𝗲𝗿 𝗶𝗻 𝗔𝗜 𝗟𝗲𝗴𝗶𝘀𝗹𝗮𝘁𝗶𝗼𝗻: The European Parliament has made history by passing the EU AI Act, a comprehensive piece of legislation set to shape the future of artificial intelligence in Europe and potentially beyond. This act heralds a new standard for AI governance, coming into force 20 days after its publication in the Official Gazzette of the European Union. 🇪🇺 𝙄𝙢𝙥𝙡𝙚𝙢𝙚𝙣𝙩𝙖𝙩𝙞𝙤𝙣 𝙋𝙝𝙖𝙨𝙚𝙨: 𝙋𝙝𝙖𝙨𝙚 1 (6 𝙈𝙤𝙣𝙩𝙝𝙨): Laying the groundwork with Chapter I (General Provisions) and Chapter II (Prohibited AI Practices). 𝙋𝙝𝙖𝙨𝙚 2 (12 𝙈𝙤𝙣𝙩𝙝𝙨): Additional provisions come into play, including Chapter III - Section 4 (Notifying Authorities and Notified Bodies), Chapter V (General-Purpose AI Models), and Chapter VII (Governance). Chapter XII (Penalities) Fines for providers of general-purpose AI models as per Article 101 will not be applied yet. 𝙋𝙝𝙖𝙨𝙚 3 (24 𝙈𝙤𝙣𝙩𝙝𝙨): The AI Act is in full swing, with all rules except penalties for general-purpose AI models becoming applicable. 𝙋𝙝𝙖𝙨𝙚 4 (36 𝙈𝙤𝙣𝙩𝙝𝙨): High-risk AI systems face stringent obligations as per Article 6(1), marking the full establishment of the Act's regulations. This groundbreaking legislation establishes a timeline for both AI developers and users to adapt to new regulations, ensuring responsible deployment and fostering an environment of trust. Let's come together to discuss how this strategic policy can steer innovation and ethical standards. How are you preparing for these changes? #AIRegulation #EUParliament🇪🇺 🇮🇹 #DigitalTransformation #ResponsibleAI

🚀 Artificial Intelligence Act: Council Approves First Worldwide Rules on #AI! Today, the Council has given the green light to the groundbreaking Artificial Intelligence Act, the first global regulation of its kind. This flagship legislation follows a ‘risk-based’ approach, ensuring that higher-risk #AIsystems are subject to stricter rules to protect society. The #AIAct aims to foster the development of safe and trustworthy AI systems within the EU's single market, balancing #innovation with the protection of fundamental rights. Key aspects include: ➡ Risk Classification: High-risk AI systems face stringent requirements, while limited-risk systems have lighter transparency obligations. ➡Banned Practices: Prohibition of AI uses like cognitive behavioural manipulation, social scoring, and predictive policing. ➡Governance: Establishment of an AI Office, scientific panel, AI Board, and advisory forum for effective enforcement. ➡Innovation Support: Regulatory sandboxes for testing innovative AI in real-world conditions. ➡Transparency and Protection: Impact assessments for high-risk AI systems and increased transparency for public entities using such systems. This landmark law sets a global standard for #AIregulation, emphasizing trust, transparency, and accountability. It's a significant step forward for Europe's #digitalfuture and innovation. #AIA #ArtificialIntelligence #EU #DigitalTransformation #Regulation https://lnkd.in/d3933wu8

EU policymakers have been trying hard to find a compromise on various topics in the AI Act (although one wonders whether such an important law needs to be progressed in 22-hour marathon work sessions while sleep-deprived!). Looking at the legal text from #AIAct trilogues on general-purpose AI models, published in Politico, there remain major gaps that mean the first comprehensive AI rulebook will fail to adequately protect EU citizens’ health, safety, and fundamental rights. Namely, there are no requirements on systemically risky model providers – of products that will be the bedrock of our digital society and economy – to allow early and priority access to their models to external vetted experts to conduct adversarial testing, including red teaming. This was in previous AI Act proposals and has been secured by the UK AI Safety Institute. Nor is there any requirement to open up these models to independent audits upon a reasoned request by the Commission or AI Office to check compliance. Given competitive commercial pressures, we cannot trust these large corporations to prioritise safety over release. While they currently perform their own internal testing, these are insufficient, as their incentives are aligned with private interests, not the general interest.

The #EuropeanUnion seems to have agreed on rules for #ArtificialIntelligence. This may have dealt a serious blow to the whole industry. Even if the aim of this new legislation is well-intentioned - who is not against discrimination - a lot of damage will be done by this restrictive regime. It is not a "launch pad ... to lead the global AI race" (Thierry Breton, EU commissioner), but potentially too high a burden for innovative companies that may now consider other locations for their business activities. It is incomprehensible why a #riskbased approach for the use of AI is not being proposed, but rather a direct shot at the technology itself. For example, the potential ban on the use of AI for scoring the economic situation is a major obstacle to the use of Machine Learning for credit scoring, if not a killer. Once more, some regulatory bureaucrats risk regulating the EU into #stagnation and economic insignificance with well-intentioned but unsuitable regulation on top of legislation such as the Data Act. It is to be hoped that the implementation of this legislation will refrain from making the worst mistakes and that the damage will be minimized. On the other hand, this development offers great opportunities for the Tech Sector in the #UK, #Norway and #Switzerland. https://lnkd.in/eJGtT5pe

Yesterday's EU Commission Decision establishing the European Artificial Intelligence Office left me very confused. First of all confused as a European citizen by the verbose and overlapping provisions of the Decision...so, the EAIO will "contribute to fostering actions and policies in the Commission that reap the societal and economic benefits of AI technologies pursuant to Article 5", but then of course it was paramount to specify that the office will "collaborate with the relevant Directorate-Generals and Services of the Commission pursuant to Article 5".. But European citizens will probably also be left scratching their heads by the juxtaposition of very broad, strategic missions (e.g. monitoring the unforeseen risks stemming from general-purpose AI models..) and very trivial secretarial tasks (providing the Secretariat for the AI Board and its subgroups, and providing administrative support to the advisory forum and the scientific panel , where applicable, including by providing the administrative set-up, organising meetings and preparing the relevant documents). As for lawyers, they may be wondering what EAIO's actual scope of action is, as it is limited by the authority and responsibilities of the member states' national supervising authorities under the AI Act..which has not been approved yet. (Some lawyers may also be left with the nagging suspicion that in article 3 of the Decision might linger a slight confusion between General-Purpose and General AI, once again). Personally, in a context in which the US have swiftly regulated certain AI issues by a Presidential Executive Order and the UK have recently signalled, again, its unwillingness to regulate AI, I wonder whether it is good publicity for the EU regulatory model to add to the still unfolding AI Act drama by setting up a highly bureaucratic body even before approving the Act. The critics of the EU rights-driven regulatory model are up in arms, and the wider AI community is listening. https://lnkd.in/dMbCyNpZ #ai #artificialintelligence #machinelearning #aiact #venturecapital #vc #regulation

🚨 𝗣𝗥𝗘𝗦𝗦 𝗥𝗘𝗟𝗘𝗔𝗦𝗘 | 𝗘𝗗𝗣𝗦 𝘀𝘁𝗮𝘁𝗲𝗺𝗲𝗻𝘁 𝗶𝗻 𝘃𝗶𝗲𝘄 𝗼𝗳 𝘁𝗵𝗲 𝟭𝟬𝘁𝗵 𝗮𝗻𝗱 𝗹𝗮𝘀𝘁 𝗣𝗹𝗲𝗻𝗮𝗿𝘆 𝗠𝗲𝗲𝘁𝗶𝗻𝗴 𝗼𝗳 𝘁𝗵𝗲 𝗖𝗼𝗺𝗺𝗶𝘁𝘁𝗲𝗲 𝗼𝗻 𝗔𝗿𝘁𝗶𝗳𝗶𝗰𝗶𝗮𝗹 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 (𝗖𝗔𝗜) 𝗼𝗳 𝘁𝗵𝗲 𝗖𝗼𝘂𝗻𝗰𝗶𝗹 𝗼𝗳 𝗘𝘂𝗿𝗼𝗽𝗲 𝗱𝗿𝗮𝗳𝘁𝗶𝗻𝗴 𝘁𝗵𝗲 𝗙𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸 𝗖𝗼𝗻𝘃𝗲𝗻𝘁𝗶𝗼𝗻 𝗼𝗻 𝗔𝗿𝘁𝗶𝗳𝗶𝗰𝗶𝗮𝗹 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲, 𝗛𝘂𝗺𝗮𝗻 𝗥𝗶𝗴𝗵𝘁𝘀, 𝗗𝗲𝗺𝗼𝗰𝗿𝗮𝗰𝘆 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗥𝘂𝗹𝗲 𝗼𝗳 𝗟𝗮𝘄 🤖🌐 "Despite the commendable efforts of the Members, Participants and Observers in CAI, the draft Framework Convention, as it stands now, could prove to be a missed opportunity to lay down a strong and effective legal framework for the development and uptake of trustworthy AI. The EDPS considers that such framework should not only be consistent with European values and the protection of individuals’ fundamental rights and freedoms, but also provide for clear and strong safeguards for the persons affected by the use of AI systems. "[T]he EDPS is concerned that the very high level of generality of the legal provisions of the draft Framework Convention, together with their largely declarative nature, would inevitably lead to divergent applications of the Convention, thus undermining legal certainty, and more generally its added value… Futhermore, any limitation of the scope of the future Framework Convention only to the activities undertaken by public authorities or entities acting on their behalf would contradict the overall policy objective of the Framework Convention, as defined by the Committee of Ministers. "Lastly, while AI systems may offer unprecedented opportunities, certain AI uses and activities may also pose very serious risks that could severely undermine human dignity and individual autonomy, human rights, democracy and the rule of law. The EDPS is deeply concerned by the absence of ‘red lines’ in the draft Framework Convention, which from the outset prohibit AI applications posing unacceptable levels of risk." 🔗 https://lnkd.in/ez75zPsT

AI Act | News from our team in #Brussels The European Parliament adopts pioneer legislation on artificial intelligence On 13th March 2024, the European Parliament has formally adopted the Artificial Intelligence Act, which marks the first binding worldwide horizontal regulation on AI, setting a common framework for the use and supply of AI systems in the EU.   The main elements of the regulation can be summarized as follows:   Rules on high-impact general-purpose AI models that can cause systemic risk in the future, as well as on high-risk AI systems A revised system of governance with some enforcement powers at EU level Extension of the list of prohibitions but with the possibility to use remote biometric identification by law enforcement authorities in public spaces, subject to safeguards Better protection of rights through the obligation for deployers of high-risk AI systems to conduct a fundamental rights impact assessment prior to putting an AI system into use.   The regulation is still subject to a final lawyer-linguist check and is expected to be finally adopted before the end of the current legislature, through the so-called corrigendum procedure. Once the Council also endorsed it, the regulation will enter into force 20 days after its publication in the EU Official Journal, and be fully applicable 24 months after its entry into force, except for:   Bans on prohibited practises – Six months after the entry into force date; Codes of practise – Nine months after the entry in force; General-purpose AI rules including governance – 12 months after entry into force; Obligations for high-risk systems – 36 months after entry into force.   Wondering how the AI Act might affect your business? Grimaldi Alliance is here to help! Get in touch to discuss its implications and how you can get prepared. Read more and download the paper 👉https://lnkd.in/eHFzK9hp #GrimaldiAlliance #GAHereThereAndEverywhere #AllOverTheWorld #AiAct

https://lnkd.in/eyhUsxqm Under the #aiact, #machinelearning systems will be divided into 4 main categories according to the potential #risk they pose to society. The systems that are considered high risk will be subject to stringent rules that will apply before they enter the EU market. It will be interesting to see how the AI Act will impact US #productdevelopment and #governance.