Danny Yu & Benjamin Cedric Larsen
Various AI regulatory regimes are currently emerging in Europe, the United States, China, and elsewhere. But what do these new regulatory regimes mean for businesses and for the adoption of self-regulatory and compliance-based tools and practices? It also outlines how it might appear to branch in some cases. Second, we discuss what this means for the company and its global operations. Third, the use of AI and regulation sits between soft law practices and new hard law measures, so we comment on the future direction as they become increasingly complex.
Conceptualizing AI governance
Two different but related forms of AI governance are emerging today. One is soft law governance, which acts as self-regulation based on non-legislative policy instruments. This group includes private companies that publish principles, guidelines, internal audit and evaluation frameworks for developing ethical AI. Actionable mechanisms by the private sector typically focus on developing concrete technical solutions, such as internal audits, standards, or developing explicit normative encodings. Soft law governance also involves multi-stakeholder organizations such as The Partnership on AI, international organizations such as the World Economic Forum, standards setting bodies such as ISO/IEC, CEN/CENELEC and NIST, and interest groups such as the Computing Society. . machinery (ACM), etc. This means that soft law governance and related mechanisms are essential to setting defaults for how AI technology is managed.
Hard law measures, on the other hand, are accompanied by laws and legally binding regulations that define what is permitted or prohibited. The regulatory approach generally refers to compliance with laws, issuance of standards-related certifications, and the creation or adaptation of laws and regulations for AI systems. Policy makers are currently considering several approaches to regulating AI. These approaches can be broadly categorized into AI-specific regulations (EU AI laws), data-related regulations (GDPR, CCPA, COPPA), and existing laws and regulations (antitrust and anti-discrimination laws). , and domain or sector specific regulations (HIPAA and SR 11-7).
New regulatory landscape
More than 200 initiatives targeting AI governance and regulation have already been published, according to the OECD AI Policy Observatory, which surveys 69 countries and territories. This effort covers a variety of areas such as antitrust concerns, interoperability standards, risk mitigation (hereafter consumer and social protection, public service delivery, protection of public value, etc.) .
Many countries have adopted national AI strategies, but not all countries and regions have adopted the same approach to AI governance and regulation. Different approaches are tied to a country’s existing institutions, including culture, values, and economic considerations such as innovation. Before you understand what this means for companies and their international operations, here are some examples of new AI regulations.
in many ways, european union We are at the forefront of data and AI regulation. The EU AI Act (“AIA”), which will come into force in phases from 2024, will establish a set of horizontal rules for the development and use of AI-powered products, services and systems within the EU. increase. The law is modeled on a risk-based approach, banning AI systems that pose unacceptable risks across the board, while high-risk systems include independent audits and new forms of monitoring and control. Subject to conformity assessment. Limited risk systems are subject to transparency obligations, and few or no risk systems are immune to EU AI law. The EU is also proposing an AI Liability Directive aimed at harmonizing national liability rules for AI.
in the England, the government released proposals to regulate the use of AI technology in June 2022. The proposal focuses on a “light-touch” sectoral approach that encourages guidance, voluntary action, and sandbox environments as a means of evaluating and testing AI technologies before they reach the market. . The proposal aims to reflect a less centralized approach than the EU AI law.
of Canada, The Directive on Automated Decision-Making came into force in April 2019 to ensure that the use of AI for administrative decision-making by governments is consistent with the core values of public administration. Canada’s Artificial Intelligence and Data Act (“AIDA”) will be introduced in June 2022 and, if approved, will be the first law in the country to regulate the use of AI systems. AIDA’s purpose is to establish common requirements across Canada for the design, development and deployment of artificial intelligence technologies consistent with national values and international standards.
usaApproaches to artificial intelligence are more fragmented and are generally characterized by the idea that companies must remain in control of industrial development and governance-related standards.
Read the full story at Competition Policy International.
