US Seeks to Eliminate the EU AI Act’s Code of Practice
The ongoing debate regarding the regulation of artificial intelligence (AI) has intensified, particularly concerning the European Union (EU) AI Act. Despite its intentions to foster a more transparent and copyright-conscious AI development landscape, critics argue that the accompanying rulebook may hinder innovation and impose excessive burdens on enterprises.
Context of the EU AI Act
The EU AI Act aims to establish a comprehensive framework for the deployment of general-purpose AI (GPAI) models, especially those that pose systemic risks. As stakeholders work on drafting the code of practice, the US government has voiced concerns over its implementation, with reports indicating that President Donald Trump is pressuring European regulators to abandon the proposed rulebook.
Critics from the US assert that the code of practice could stifle innovation and extend the reach of the existing AI law by introducing new, unnecessary compliance requirements. The Mission to the EU has actively reached out to European authorities to express opposition to the rulebook in its current form, emphasizing the potentially burdensome nature of its obligations.
Shifting Responsibilities in AI Compliance
The European Commission has described the code of practice as a vital tool for AI providers aiming to demonstrate compliance with the EU AI Act. Although voluntary, the code is intended to assist organizations in meeting regulations related to transparency, copyright, and risk mitigation.
The drafting process involves a diverse group of stakeholders, including AI model providers, industry organizations, and civil society representatives, overseen by the European AI Office. The deadline for the completion of this code is set for the end of April, with the final version scheduled for presentation to EU representatives in May.
Implications for Enterprises
As the regulatory landscape evolves, the responsibility for ensuring responsible AI practices is shifting from vendors to the organizations deploying AI technologies. Experts indicate that any business operating within Europe must develop its own AI risk playbooks, which should include measures such as privacy impact assessments, provenance logs, and red-team testing. This proactive approach is essential to mitigate potential contractual, regulatory, and reputational risks.
Non-compliance Risks
The consequences of non-compliance with the EU AI Act could be severe, including fines that could reach up to 7% of global revenue. This underlines the importance of organizations adapting to the evolving compliance landscape, regardless of the eventual outcome of the EU’s deliberations on the code of practice.
Potential for a Lighter Regulatory Touch
If the US administration’s approach to AI legislation gains traction globally, it could result in a more lenient regulatory environment with diminished federal oversight. Recent actions, including Executive Order 14179, indicate a shift towards reducing barriers to American leadership in AI, with new guidelines emphasizing economic competitiveness over stringent regulatory measures.
As the landscape of AI regulation continues to change, both domestic and international stakeholders must navigate the complexities of compliance while fostering innovation in AI technologies.
