OECD members: Issued report on the pilot phase of the application of Hiroshima Process International Code of Conduct for organisations developing advanced AI systems

On 23 December 2024, the OECD Secretariat published a report on the pilot phase of the application of the Hiroshima Process International Code of Conduct (Code of Conduct) for organisations developing advanced AI systems. The pilot phase was conducted from 19 July 2024 to 6 September 2024 and gathered responses from 20 organisations across 10 countries. The Code of Conduct employs a risk-based approach to encourage safe, secure, and trustworthy Artificial Intelligence (AI) on a global scale. It provides voluntary guidance for organisations involved in the development of advanced AI systems, including foundation models and generative AI. The Code of Conduct consists of high-level principles rather than specific requirements, and was commended by participants for its ability to address critical AI development and risk management areas while fostering alignment with international standards. However, areas for improvement were identified, such as consolidating repetitive questions, enhancing formatting options, clarifying instructions, explaining key terms, and ensuring alignment with other voluntary reporting mechanisms. Concerns were also raised about the use and sharing of responses, particularly regarding the confidentiality of sensitive information. The G7 plans to incorporate this feedback into a revised framework and will continue engaging stakeholders for ongoing updates and implementation support.