Every AI Policy Ever · open in the interactive register
Published (non-binding, current)Non-binding · Strategy / action plan
Record INT-20
Voluntary action plan agreed at the second in-person meeting of the Hiroshima AI Process Friends Group (Tokyo, 15-16 March 2026) to promote safe, secure and trustworthy AI over the following year. It lists workshops on the OECD AI Principles toolkit and the HAIP Reporting Framework, capacity-building and information-sharing on AI governance and transparency, a group study on interoperability between the HAIP and other frameworks, AI Academy courses with the World Bank and talent programmes for the Global South. An annex details member and partner projects, including a Singapore IMDA workshop on chatbot safety transparency and agentic-AI risks.
Borderline scope Voluntary action plan of a government group; safety content is procedural (reporting, transparency, interoperability). Core enough under the HAIP precedents, but a scope note would be reasonable.
The Friends Group was launched in May 2024 as a voluntary framework of countries supporting the spirit of the HAIP; as of March 2026 it had 66 countries and regions and 38 organisations in its Partners Community. The Tokyo meeting of 15-16 March 2026, hosted by Japan's Ministry of Internal Affairs and Communications with the OECD and the World Bank as co-hosts, was attended by 34 Friends Group countries and regions, Brazil as an observer and 32 partner organisations. The plan is to be shared in coming international discussions. At the meeting the OECD announced a pilot version aimed at revising the HAIP Reporting Framework (OECD-02), and the World Bank announced an AI Academy to support capacity building in the Global South. The G7 Digital and Technology Ministers, meeting in Paris on 29 May under France's presidency, welcomed the Friends Group's initiatives and 'the related Action Plan announced in March 2026' (G7-04).
Open the official source https://www.soumu.go.jp/main_sosiki/joho_tsusin/eng/pressrelease/2026/pdf/image_20260316_2_2.pdf
Secondary source: https://www.soumu.go.jp/main_sosiki/joho_tsusin/eng/pressrelease/2026/3/16_2.html
Archived. Last attempt 2026-10-17 16:30:08 UTC
G7 Hiroshima Code of Conduct G7 · Published (non-binding, current)
HAIP Comprehensive Policy Framework G7 · Published (non-binding, current)
HAIP Reporting Framework OECD / G7 · Published (non-binding, current)
G7 Digital & Tech Ministerial Declaration 2026 (Paris) G7 · Published (non-binding, current)
OECD AI Principles OECD · Published (non-binding, current)
Hiroshima AI Process Friends Group (66 countries and regions) with its Partners Community, at the Second In-Person Meeting hosted by Japan's Ministry of Internal Affairs and Communications in Tokyo. Hiroshima AI Process Friends Group Action Plan 2026. 2026-03-16. https://www.soumu.go.jp/main_sosiki/joho_tsusin/eng/pressrelease/2026/pdf/image_20260316_2_2.pdf. Status: Published (non-binding, current); verified 2026-10-17 (level A). Local copy archive/INT-20/2026-10-11/official.pdf retrieved 2026-10-11 19:18:53 UTC, SHA-256 4c324616e053f45786b3de1dd70976380d7df8667034b9445cc2f9d046cf75ac. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record INT-20.
Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.