Unpacking the EU AI Act Code of Practice with Marietje Schaake

Marietje Schaake's Path to AI Policy

• 💡 Marietje Schaake's early career was driven by a curiosity about technology's role in societal change and its intersection with politics.
• 📌 Her experience in the European Parliament, particularly during the Arab uprisings, highlighted the dual nature of technology, including the use of sophisticated surveillance tools by dictatorships.
• ✅ Schaake's core motivation is to ensure that fundamental values like rule of law, democracy, and human rights are the guiding principles for technology development and regulation.

The EU AI Act and Code of Practice Explained

• 🔑 The EU AI Act is a landmark legislation focused on establishing a conceptual framework for assessing and mitigating AI risk.
• 🎯 The Code of Practice for General Purpose AI Models serves as a crucial roadmap, helping companies demonstrate compliance with the AI Act, especially for models posing systemic risks.
• 📊 While the AI Act is legally binding, the Code of Practice offers a series of steps and commitments that companies can adhere to, fostering rigorous documentation and information sharing.

Addressing Systemic Risks in AI

• ⚠️ The Code of Practice primarily targets systemic risks to areas like national security (e.g., bioweapons, cyber weapons) and public health.
• 🔬 A key concern is that AI could make weapons of mass destruction accessible to a broader range of actors, moving beyond state-level capabilities.
• 📈 Companies are expected to minimize the potential for AI to enable harmful activities, recognizing that even one bad actor can erode trust in AI as a whole.

Regulatory Challenges and Approach

• 🚀 The EU's regulatory approach focuses on upstream providers—the handful of companies with unique capabilities to develop advanced AI models—due to their resources and insights.
• 🧩 A significant challenge is balancing the need for specific legal requirements with the inherent unpredictability of rapidly evolving AI technology.
• 🧠 Existing legislative and enforcement models may not be fully equipped to handle the unforeseen changes and uncertainties that AI introduces.

Measuring Success and Building Trust

• 🤝 Success of the Code of Practice will be measured by its ability to create a more agile and predictable enforcement process for both companies and regulators.
• 👏 The hope is that companies will identify and mitigate risks they might have otherwise missed, fostering trusted relationships with regulators and the public.
• ✨ Ultimately, the goal is to build trust in AI as a technology, ensuring its benefits materialize while safeguarding against potential harms and promoting predictability for all stakeholders.