The Australian government has published an interim response outlining their plans to regulate high-risk AI systems in the country. The response is guided by key principles including a risk-based approach, collaboration and transparency, and a community-centric approach. Specific measures proposed include mandatory guardrails, testing and transparency initiatives, an AI safety standard, and funding for AI initiatives to support adoption and development. The government aims to strike a balance between fostering innovation and protecting community interests, particularly privacy and security, while addressing potential harms caused by high-risk AI systems. The response reflects Australia's commitment to responsible AI practices and international cooperation.

The European Commission has announced the creation of the European Artificial Intelligence Office (AI Office), a key part of the forthcoming AI Act. The office will contribute to the implementation and enforcement of the act, and will sit within the Commission's DG CNECT department. The AI Office will be financed by the Digital Europe Programme. The EU is expected to promote early voluntary compliance with the AI Act through the Commission and the AI Office. The act is likely to come into force in the coming months.

The Federal Artificial Intelligence Risk Management Act of 2024 has been introduced in the US Congress, requiring federal agencies to comply with the Artificial Intelligence Risk Management Framework developed by the National Institute of Standards and Technology. The framework, designed to help organizations prevent, detect, mitigate, and manage AI risks, sets out four key processes including mapping, measuring, managing, and governing. The Act also includes guidance for agencies on incorporating the AI RMF, reporting requirements, and regulations on AI acquisition. Compliance with NIST’s AI Risk Management Framework may soon become a legal requirement, as several states and federal laws already draw on it.

The International Standardisation Organisation (ISO) and the International Electrotechnical Commission (IEC) have published a voluntary standard, ISO/IEC 42001, that establishes a process for implementing AI Management Systems (AIMS) in organisations. ISO/IEC 42001 is the first process standard to outline a comprehensive governance framework for responsible and trustworthy deployment of AI systems. The standard is scalable and offers auditability, providing a foundation for external entities to certify and audit AI systems in line with the risk assessment framework. Policymakers behind the EU AI Act and legislative efforts on AI in the US may adopt the standard’s objectives and modalities to guide their regional standardisation efforts. Holistic AI, a regulatory compliance solutions provider, offers services to assist organisations in complying with applicable AI regulations and industry standards.

European co-legislators have agreed a provisional agreement on the EU AI Act which seeks to harmonise AI regulations across the EU. The agreement balances innovation with fundamental rights and safety and positions the EU as a leader in digital regulation. The EU AI Act includes prohibitions on biometric identification, updated requirements for high-risk AI systems, and a two-tiered approach for general purpose AI systems. It is expected to gradually enter into force with provisions on prohibitions coming into effect within six months and provisions on transparency and governance coming into effect after twelve months. The EU AI Act will likely set a global benchmark for the ethical and responsible design, development, and deployment of AI systems.