Navigating the complex world of cryptocurrency and blockchain is already challenging enough, but now a new layer of tech regulation is emerging that could impact the future landscape: Artificial Intelligence. For those watching the intersection of crypto and AI, the latest developments in the EU’s AI Act are crucial. A new draft code of practice hints at a potentially softer approach to regulating powerful AI models, a move that could have ripple effects across tech and finance. Let’s dive into what this means for the future of AI and its implications for the crypto space. Decoding the Latest EU AI Act Draft for GPAI Models The European Union is forging ahead with its EU AI Act , a groundbreaking piece of legislation designed to govern the rapidly evolving field of artificial intelligence. A key component of this act is the Code of Practice for General Purpose AI (GPAI) models, particularly targeting the ‘Big AI’ players. The third draft of this code, released recently, suggests a move towards ‘gentler guidance,’ a shift from potentially stricter regulations. This revision aims to clarify how GPAI models can comply with the EU AI Act and avoid hefty penalties, which could reach up to 3% of global annual turnover for non-compliance. Here’s a breakdown of what’s new in this draft: Streamlined Structure: The latest draft boasts a more organized structure with refined commitments, making it easier for AI model makers to navigate and understand the requirements. Feedback Incorporation: This revision reflects feedback from previous drafts, indicating an effort to create practical and workable guidelines. Focus on Clarity: Experts are aiming for greater clarity and coherence in the final version, addressing concerns about ambiguity in earlier drafts. Key Sections: The draft is divided into sections covering commitments for GPAI models , detailed guidance on transparency and copyright, and safety/security obligations for the most powerful models. Transparency and Copyright: Wiggle Room for Big AI? One of the most closely watched areas is how the EU AI Act will address transparency and copyright issues related to Big AI . The current draft uses terms like “best efforts,” “reasonable measures,” and “appropriate measures” when discussing compliance. This language suggests a potentially more flexible approach, especially regarding data mining for model training and mitigating copyright infringement risks. Copyright Compliance – Key Changes: Aspect Previous Draft Current Draft Point of Contact for Rights Holders Required single point of contact for direct and rapid communication. Designated point of contact with easily accessible information. Handling Complaints Implied obligation to act on complaints. Possibility to refuse action on complaints deemed “manifestly unfounded or excessive.” This shift in language could be interpreted as offering Big AI companies more leeway. The removal of the explicit requirement for a “single point of contact” for rights holders and the introduction of the ability to dismiss complaints as “excessive” raises questions about the robustness of copyright protection under the new code. Safety and Security: Narrowing the Scope of Systemic Risk The EU AI Act also addresses safety and security risks associated with the most powerful AI models . Requirements to evaluate and mitigate systemic risks are specifically for models trained with significant computing power. Interestingly, the latest draft further narrows some previously recommended safety measures, reflecting feedback and potentially easing the burden on AI model makers . US Pressure and the Global AI Landscape While not explicitly mentioned in the EU’s official statements, external pressures, particularly from the US, are likely influencing the direction of AI regulation . The US administration has voiced concerns about overregulation stifling AI innovation and economic opportunity. This external pressure, combined with internal lobbying from Big AI companies, could be contributing to the observed shift towards a gentler approach in the EU’s draft code. The cancellation of the AI Liability Directive and the push for simplifying existing rules further suggest a broader trend towards deregulation within the EU. French GPAI model maker Mistral, a vocal critic of the EU AI Act , has expressed difficulties in complying with certain regulations, adding to the pressure for a more lenient framework. What’s Next? Clarifying Guidance and the Role of the AI Office The European Commission, through its AI Office, is working in parallel to provide “clarifying” guidance on the EU AI Act . This includes defining GPAI models and their responsibilities, which will be crucial in shaping how the law is applied in practice. These clarifications, expected “in due time,” could offer a pathway for EU lawmakers to balance AI regulation with concerns about stifling innovation and competitiveness, particularly in response to US lobbying efforts. Is Gentler Guidance the Right Path Forward for AI Regulation? The latest draft of the EU AI Act ’s Code of Practice for GPAI models presents a complex picture. While the streamlined structure and incorporation of feedback are positive steps towards creating workable guidelines, the softened language around copyright and the narrowing of safety measures raise questions about the level of AI regulation the EU ultimately intends to enforce. For the cryptocurrency and broader tech world, these developments are vital to monitor. The balance between fostering innovation and mitigating the risks of powerful AI will shape the future of technology and its integration into various sectors, including blockchain and digital assets. Will this gentler approach truly encourage responsible AI development, or will it inadvertently create loopholes for Big AI to operate with fewer constraints? The final version of the Code of Practice and the forthcoming clarifications from the AI Office will provide crucial answers. To learn more about the latest AI regulation trends, explore our article on key developments shaping AI features.
