G7 Hiroshima AI Process (HAIP) Reporting Framework

Ways to reach them

HAIP transparency reporting portal (for organisations that develop AI) · Written submission

https://transparency.oecd.ai/

Verified · 2026-09-22

AI-developing organisations of any size, voluntary

Not a channel for the general public; it is a self-disclosure form for developers.

HAIP overview and framework documents · Web form

https://oecd.ai/en/hiroshima

Verified · 2026-09-22

Reference

Homepage · Homepage

https://oecd.ai/en/transparency/overview

Unchecked

What it does

Voluntary transparency reporting against the G7 Hiroshima Code of Conduct for organisations developing advanced AI. Launched by the OECD Feb 2025; first reports published April 2025; a streamlined version for SMEs launched May 2026. Companies self-report on risk identification, incident handling, security, content authentication and governance; reports are published on a public portal.

Honest assessment

Useful as a source of citable company commitments you can hold firms to (as the Midas Project and SaferAI do), rather than as a place to send a concern. There is no public comment or challenge mechanism against a company's self-report — that gap is exactly what the independent trackers fill.

Concerns it covers

Frontier-developer risk management — misuse, security, and safety-framework adequacy. Self-reported, so it measures disclosure, not actual safety.

Government channel