Global: Australia Advances AI Regulation with New Expert Group, While EU Nears Adoption of AI Act

Australia has taken a significant step in the direction of regulating artificial intelligence by forming a new AI Expert Group. This group, comprising scientists, legal experts, and professionals in ethics and governance, will advise the Department of Industry, Science, and Resources on establishing guidelines for high-risk AI systems. The goal is to ensure that AI systems are transparent, well-tested, and accountable. Minister for Industry and Science Ed Husic emphasized the need for transparency and rigorous testing for sophisticated models underpinning high-risk AI systems.

This move reflects Australia’s ongoing efforts to potentially introduce mandatory guidelines for AI applications. The initiative was initially announced in September, followed by the government’s interim response to the Safe and Responsible AI in Australia consultation in January. In addition to creating the AI Expert Group, the response outlined plans to collaborate with the industry to develop a voluntary AI Safety Standard and voluntary labeling and watermarking of AI-generated materials.

The AI Expert Group, established in early February, is scheduled to work until June 30, 2024. The Australian government is currently contemplating long-term responses to AI regulation. Notable figures in the 12-member group include Bronwyn Fox, Chief Scientist at the Commonwealth Scientific and Industrial Research Organisation (CSIRO), and Aurélie Jacquet, Chair of Australia’s national AI standards committee.

Meanwhile, the European Union (EU) has moved another step closer to the adoption of the AI Act. Members of the European Parliament (MEPs) endorsed the provisional agreement on the Act at the committee level on Tuesday. The deal on technical details, approved by the EU Internal Market and Civil Liberties Committees, opens the door to formal adoption by the European Parliament and final Council endorsement. The legislation is expected to be fully applicable 24 months after entering into force, with final steps anticipated to be completed in April.

The AI Act addresses certain AI applications that threaten citizens’ rights, prohibiting activities such as biometric categorization and untargeted scraping of facial images. However, exemptions in the legislation, especially regarding emotion recognition, have drawn criticism. While the AI Act bans the use of emotion recognition in workplaces and schools, concerns persist that it may still be permitted in law enforcement, particularly for migration control. The legislation also introduces regulatory sandboxes and real-world testing at the national level, allowing companies to train AI before market placement.