Global regulation of AI

Nations and economic blocs around the world are at different stages of regulating artificial intelligence (AI), ranging from the relatively unregulated environment of the United States to the complex frameworks in the European Union.

Ahead of the February 10-11 Paris AI summit, here are key points about AI regulation in major regions:

US

Last month, President Donald Trump reversed President Joe Biden's October 2023 executive order on AI oversight. The order, which was largely voluntary, required major AI developers like OpenAI to share safety assessments and important information with the federal government. Supported by major tech companies, it aimed to protect privacy, prevent civil rights violations, and introduce safeguards on national security.

Currently, the United States has no formal AI regulations, although some privacy protections still apply. Under Trump, the country is described as returning to a "Wild West" approach to AI, with little oversight. "We’re setting all our algorithms running and going for it," said Yael Cohen-Hadria, a digital lawyer at EY.

Beijing (China)

China's government is still working on a formal law for generative AI. In the meantime, it has introduced "Interim Measures" that require AI to respect personal and business interests, ensure consent for the use of personal information, label AI-generated content, and protect users’ health. AI must also align with "core socialist values," and any content threatening the Communist Party or national security is strictly prohibited.

China's regulation focuses heavily on businesses, particularly foreign ones, while the government is likely to grant itself exceptions to its own rules.

EU

The European Union takes a different approach, with a strong emphasis on citizen protection. The "AI Act," which passed in March 2024, is the world’s most comprehensive AI regulation and takes effect this week. The law bans AI systems used for predictive policing based on profiling or biometric data that infer race, religion, or sexual orientation.

The EU follows a risk-based approach: high-risk AI systems face stricter obligations. This regulation aims to create clarity for businesses while ensuring robust protections for intellectual property, data circulation, and citizens' control over their data. "If I can access a lot of data easily, I can create better things faster," said Cohen-Hadria.

India

India, co-host of the upcoming summit, has a law regulating personal data but no specific AI laws. Existing legislation on defamation, privacy, copyright infringement, and cybercrime is being used to address harm caused by generative AI.

India's government has not yet implemented concrete AI regulations, though it acknowledges the value of its growing high-tech sector. Recent reports of government statements and an "advisory" from the IT ministry that required firms to seek government approval before deploying AI models have sparked criticism from major AI firms like Perplexity.

UK

The UK government, led by the centre-left Labour Party, has placed AI on its agenda to boost economic growth. The country, with the third-largest AI sector globally, is aiming to chart its own path in AI regulation. Prime Minister Keir Starmer’s "AI opportunities action plan," introduced in January 2024, calls for AI testing before regulation is implemented. Starmer emphasized that well-designed regulation could foster rapid and safe development of AI, while ineffective regulation might hinder its adoption, especially in key sectors. A consultation is also underway to clarify how copyright law applies to AI, focusing on protecting the creative industry.