Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems

• Authority: Other multinational
• Jurisdiction: Multinational
• Collections: Multinational;Editors' Picks
• Most recent activity: Enacted
• Most recent activity date: 2023-10-30

Risk subdomain coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the risk subdomain are explicitly described.

• Compromise of privacy
• AI system security vulnerabilities
• Disinformation, surveillance, and influence at scale
• Cyberattacks, weapon development or use
• Governance failure
• AI pursuing its own goals
• AI possessing dangerous capabilities
• Lack of capability or robustness
• Lack of transparency or interpretability

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the risk subdomain are described.

• Unfair discrimination and misrepresentation
• Unequal performance across groups
• False or misleading information
• Pollution of information ecosystem
• Fraud, scams, and targeted manipulation
• Overreliance and unsafe use
• Competitive dynamics
• Environmental harm
• Multi-agent risks

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document has good coverage of approximately 15-17 subdomains, with strong focus on malicious actors (4.1, 4.2, 4.3), AI system security (2.2), privacy compromise (2.1), competitive dynamics (6.4), governance failure (6.5), and AI safety failures (7.1, 7.2, 7.3, 7.4). Coverage is concentrated in security, misuse prevention, AI safety, and governance domains, with minimal coverage of discrimination/toxicity and human-computer interaction risks.

Sector coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the sector are explicitly described.

• Information
• Scientific Research & Development Services

All other sectors had no coverage in this document.

Sectors Summary

This document does not govern AI use within specific economic sectors. Rather, it governs organizations that develop advanced AI systems (AI developers), which primarily operate in the Information and Scientific Research and Development Services sectors. The document is technology-developer focused, not sector-application focused.

Legislative Status

Soft Law

Legislative Status Summary

This is a voluntary international code of conduct with no binding legal obligations or enforcement mechanisms. The document explicitly states it provides 'voluntary guidance' and relies on organizational commitment rather than legal compulsion.

Actors

Proposer

• Entity Types: AI_Governance_Actor
• Summary: The document is produced by the Hiroshima Process, building on OECD AI Principles, with commitment to develop proposals in consultation with OECD and GPAI. The Hiroshima Process is a G7-led international initiative.

Target

• Entity Types: AI_Developer
• Summary: The document explicitly targets organizations developing the most advanced AI systems, including foundation models and generative AI. It states that endorsers may include entities from academia, civil society, private sector, and public sector.

Enforcer

• Summary: The document does not designate any enforcement body. It is a voluntary code of conduct with no formal enforcement mechanisms. Organizations are expected to self-regulate through internal governance structures.

Monitor

• Entity Types: AI_Governance_Actor
• Summary: The document commits to developing monitoring tools and mechanisms in consultation with OECD and GPAI. Organizations are encouraged to implement self-assessment mechanisms. However, formal monitoring bodies are not yet established.

AI Lifecycle Stages

Lifecycle stages covered by the documents, as defined in the OECD Framework for the Classification of AI Systems

• Plan and Design
• Collect and Process Data
• Build and Use Model
• Verify and Validate
• Deploy
• Operate and Monitor

AI Lifecycle Summary

The document explicitly covers all stages of the AI lifecycle, with particular emphasis on development, testing, deployment, and post-deployment monitoring. It repeatedly references 'across the AI lifecycle' and 'throughout the AI lifecycle' as a core principle.

Technical Scope

Terms relating to technical scope that are mentioned or implicitly referenced in the document

• AI Models
• Foundation Models
• AI Systems
• Generative AI

Technical Scope Summary

The document explicitly focuses on 'advanced AI systems, including the most advanced foundation models and generative AI systems'. It does not use compute thresholds or explicitly distinguish between general purpose and task-specific AI. It does not explicitly mention open-weight models, though it addresses securing model weights.