Skip to content

Akiko Murakami Named First Director of Japan's AI Safety Institute

Akiko Murakami, Sompo Japan CDO and a Watson veteran, is the first director of Japan's AI Safety Institute. Here is what the appointment means for AI policy and business.

Akiko Murakami Named First Director of Japan's AI Safety Institute

Japan’s AI policy architecture acquired a named face on February 1, 2024, when Deputy Chief Cabinet Secretary Hideki Murai announced that Akiko Murakami would be the first director of the AI Safety Institute (AISI). Two years on, AISI has become a central fixture in Japan’s AI policy framework, and Murakami’s appointment remains the decision that set its commercial and technical tone. For foreign AI developers, enterprise adopters, and multinational technology firms, the character of AISI under its founding director directly shapes how model evaluations, public-sector procurement, and AI Safety rules are designed in Japan.

Akiko Murakami Named First Director of Japan's AI Safety Institute

Context: why Japan created an AI Safety Institute

AISI was established in early 2024 to give Japan a technical counterpart to the UK AI Safety Institute and the US AI Safety Institute, creating an international network for coordinated evaluation of frontier AI systems. Its creation followed commitments made under the Hiroshima AI Process during Japan’s G7 presidency and the growing domestic recognition that principled guidelines alone would not be sufficient as generative AI moved into enterprise and critical-infrastructure use.

The institute was placed under the Information-technology Promotion Agency (IPA), which sits under METI jurisdiction. That institutional choice matters: IPA already houses cybersecurity capability and technical evaluation infrastructure, and its METI anchoring positions AISI within the industrial policy apparatus rather than as a pure regulator. The practical effect is that AISI’s work is designed to be useful to Japanese industry as well as to policymakers.

What happened

Deputy Chief Cabinet Secretary Murai announced the appointment at a February 1, 2024 press conference. Murakami, then serving as Chief Digital Officer of Sompo Japan Insurance, was selected for her combination of deep technical AI experience and enterprise deployment expertise. She previously played a key role in the development of IBM Japan’s Watson AI system, giving her first-hand exposure to the governance and safety issues that arise when commercial AI systems meet regulated industries.

Murai described Murakami as the ideal candidate to navigate AI safety at a global scale, signalling that the government viewed AISI from the outset as an internationally-facing body. That framing has held.

The announcement also confirmed AISI’s operational placement under IPA and METI jurisdiction, confirming that the institute would work closely with:

  • METI’s industrial policy and AI governance teams
  • The Cabinet Office AI strategy apparatus
  • The Personal Information Protection Commission on data issues
  • International counterparts in the UK, US, and other AISI Network members

Commercial impact

Two years into operation, AISI’s impact on the commercial environment for AI in Japan is increasingly concrete.

Model evaluation is becoming a market signal

AISI has published guidance on red-teaming, evaluation methodologies, and safety benchmarks adapted to Japanese language and use cases. Enterprise procurement teams, particularly in regulated sectors, increasingly reference AISI outputs when assessing AI vendors. Vendors that have engaged with AISI evaluations have a credibility advantage in Japan public-sector and regulated-industry sales.

Public-sector procurement is tightening

Ministries and agencies acquiring AI systems increasingly look for alignment with AISI guidance and the AI Guidelines for Business issued by METI and MIC. Foreign vendors competing for Japanese government work should expect documentation requirements on training data, evaluation, red-teaming, and incident response.

Enterprise governance is anchoring on Japanese standards

Japanese enterprises, particularly in finance, healthcare, and critical infrastructure, are using AISI outputs and the AI Guidelines for Business as internal governance anchors. Multinational vendors selling into these accounts need their global responsible-AI frameworks to map cleanly onto Japanese expectations.

International coordination matters

AISI’s membership in the International AISI Network means its evaluation frameworks are increasingly interoperable with UK and US counterparts. That reduces duplicative work for multinationals but also creates shared expectations on model transparency, safety testing, and incident reporting.

What is next

Several threads are worth watching closely.

The AI Basic Law

Japan’s emerging AI Basic Law and follow-on implementing measures will define AISI’s statutory footing and authorities. Expect the institute’s role in pre-deployment evaluation, incident response, and regulatory advice to expand.

Frontier model evaluation

AISI is building capacity for pre-deployment evaluation of frontier models, in coordination with international partners. Foreign developers of capable models should plan for voluntary (and potentially mandatory) engagement.

Sector-specific AI safety

Expect AISI outputs to feed into sector-specific AI rules in finance (FSA), healthcare (MHLW/PMDA), transportation (MLIT), and critical infrastructure. Sector regulators will rely on AISI technical work to calibrate their rules.

Talent and capacity

AISI is building out its technical and policy teams. Its credibility, and its leverage over industry, scales with who it hires and who it partners with. Industry engagement at the working-group level is where this gets shaped.

Why this matters for public affairs in Japan

AISI sits at the intersection of industrial policy, safety regulation, and international AI diplomacy. Foreign AI developers and enterprise users who engage AISI substantively, through technical contributions, joint evaluations, and participation in industry panels, will have a meaningful voice in how Japan’s AI rules are operationalised. Those who do not will inherit frameworks designed without their input.

Gemini Group advises multinational AI developers, enterprise adopters, and technology firms on Japan AI policy, AISI engagement, METI and Cabinet Office coordination, and the AI Basic Law process. Contact us to discuss your Japan AI public affairs strategy.