Countries will set their rules as AI gathers pace

As artificial intelligence (AI) advances, countries are racing to establish their own regulations, potentially fragmenting the global digital market. Concerns about AI risks, including misinformation, job displacement, and bias, are driving this push for rules. This article outlines the current state of AI regulation, focusing on the approaches of the US, EU, and China, recognising that this landscape is rapidly evolving.

EU leading global tech regulation

The European Union (EU) has taken the lead in digital regulation. The EU Parliament has passed the AI Act, currently under discussion with the European Council and Commission. It categorises AI into different risk levels, banning high-risk applications like social scoring. High-risk AI requires registration and conformity declarations. Originally designed for single-task AI, this framework adapts to generative AI and foundation models. All AI is considered high risk with an emphasis on transparency and liability. The final AI Act text is expected by late 2023 or early 2024.

Additional EU regulations will shape AI. The AI Liability Directive shifts the burden to companies for algorithm-induced damages. The Data Act focuses on fair data access, and the Data Governance Act enables data sharing. GDPR, effective from 2018, guards personal data privacy concerning AI. The Digital Markets Act aims to ensure competition, targeting major cloud providers.

Challenges in the US

The US prioritises innovation over regulation and prefers market-driven self-regulation. Intense tech rivalry with China leads to strict controls, like those on semiconductors.

US political challenges hinder regulations. Legislative proposals like the SAFE innovation framework and the Algorithmic Accountability Act face uncertain passage before the 2024 term end, given previous tech legislation stagnation. The executive branch uses existing authority for new AI use cases. The White House introduced an AI Bill of Rights, secured tech company commitments, and various agencies issued a joint statement. The FTC investigates OpenAI. Judicial decisions on internet company immunity are expected.

China’s approach

China actively pursues AI regulation. Comprehensive AI regulations are expected soon. China aims to lead in AI globally while preventing alternative centers of power. Early regulations focus on recommendation algorithms and greater tech scrutiny.

China’s draft regulation on generative AI emphasizes socialist values, accuracy, and preventing fake content. Discussions center on registering models, algorithms, and scrutinising training data. These rules emphasise responsibility, privacy, accuracy, and misinformation, but their global impact may be limited.

Global implications and challenges

While other countries introduce or consider AI regulations, the EU’s potential stringency stands out. It seeks global influence in AI regulation as it did with GDPR. However, many countries view AI as a competitive advantage and favour innovation over strict regulation, potentially disadvantageous to the EU. AI regulations worldwide contribute to increased industry fragmentation and regionalisation, a growing trend.

The analysis and forecasts featured in this piece can be found in EIU’s Country Analysis service.