---
title: Japan Trails on Generative AI Policy, Mostly by Not Deciding
description: "Japan's 2026 ICT White Paper: 68.9% of Japanese firms have a generative AI policy vs 82.8% in the US. Only 3.5% ban it; 20.2% haven't decided yet."
image: https://nobusato.jp/hubfs/japan-generative-ai-adoption-gap-en.png
---

[Nobu Sato](https://nobusato.jp/en?hsLang=en)

[Home](https://nobusato.jp/en?hsLang=en) [Profile](https://nobusato.jp/en/profile?hsLang=en) [Articles](https://nobusato.jp/en/blog?hsLang=en) Guides

[Launching in Japan](https://nobusato.jp/en/blog/guide-launching-in-japan?hsLang=en)[Organizations & Hiring](https://nobusato.jp/en/blog/guide-organizations-and-hiring?hsLang=en)[Japan by the Numbers](https://nobusato.jp/en/blog/guide-japan-by-the-numbers?hsLang=en)[Career](https://nobusato.jp/en/blog/guide-career?hsLang=en)

[Glossary](https://nobusato.jp/en/blog/glossary?hsLang=en) [Contact](https://nobusato.jp/en/contact?hsLang=en)

[日本語](https://nobusato.jp/ja/blog/japan-generative-ai-adoption-gap)[English](https://nobusato.jp/en/blog/japan-generative-ai-adoption-gap)[中文](https://nobusato.jp/zh/blog/japan-generative-ai-adoption-gap)

[Japan by the Numbers](https://nobusato.jp/en/blog/tag/japan-by-the-numbers)

# Japan Trails on Generative AI Policy, Mostly by Not Deciding

[Nobu Sato](https://nobusato.jp/en/profile?hsLang=en)

![Bar chart of firms with a generative AI use policy: Japan 68.9%, US 82.8%, Germany 82.2%, China 95.4%](https://nobusato.jp/hs-fs/hubfs/japan-generative-ai-adoption-gap-en.png?width=1200&height=630&name=japan-generative-ai-adoption-gap-en.png)

If you sell AI tools into Japan, your toughest competitor is probably a company that hasn't made up its mind. In the Japanese government's latest survey, 20.2% of Japanese firms said they have no clear policy on generative AI. In the US the figure was 7.8%. In China, 2.3%.

The headline number is the one most people quote. According to the 2026 White Paper on Information and Communications from Japan's Ministry of Internal Affairs and Communications (MIC), published in July 2026, 68.9% of Japanese companies have a policy to use generative AI, either actively or in limited areas. The US is at 82.8%, Germany 82.2%, China 95.4%. Japan comes last. But look at where the missing third sits, and the story changes.

## Japanese firms aren't banning AI. They're undecided.

Only 3.5% of Japanese companies in the survey ban generative AI. That's lower than the US (9.1%) and Germany (7.8%). What Japan has instead is a large middle: 20.2% with no clear policy, plus 7.4% who answered "don't know." Together that's 27.6%. The comparable figure is 8.1% in the US, 10.0% in Germany and 2.6% in China.

So most of the gap with other countries is firms that haven't decided yet. That matters if you're building a go-to-market plan. A buyer who hasn't decided and a buyer who has said no need very different conversations.

Use activelyUse in limited areasBannedNo clear policyDon't knowJapan30.138.820.27.4policy 68.9%US34.648.29.17.8policy 82.8%Germany40.541.77.89.7policy 82.2%China49.845.6policy 95.4%0%50%100%

**Fig.1**Generative AI policy by country. Japan stands out for firms with no clear policy (20.2%) and don't know (7.4%), while its ban rate (3.5%) is lower than the US or Germany. Source: MIC, 2026 White Paper on Information and Communications, Fig. I-2-1-1 (company survey fielded January to February 2026).

## An online panel of managers at firms already digitalizing

The numbers come from an online survey of managers, run through a research firm's registered panel. Keep that in mind when you decide how far to trust them.

MIC commissioned the study and fielded it in January and February 2026. It got 515 valid responses in Japan and 309 each in the US, Germany and China. To qualify, a respondent had to be a manager or above who understands company-wide strategy, at a firm with 10 or more employees that had started digitalization work by 2025.

Three caveats follow from that design. First, firms that haven't started any digitalization are excluded from the start, so 68.9% is not the adoption rate of Japanese companies overall. My guess is the true figure across all firms is lower, though this survey can't show it. Second, the samples aren't built the same way. In China, 250 of the 309 firms are large companies. In Japan it's 353 of 515. Within Japan, 74.0% of large firms have a policy against 58.1% of small and mid-sized ones, so part of the country gap may just reflect who got sampled. Third, it's self-reported. A manager's view of the company's policy can be wrong, and how often people pick "don't know" depends partly on habits of answering. Japanese respondents lean toward neutral options, which I looked at in [a piece on neutral survey answers in Japan](https://nobusato.jp/en/blog/japan-neutral-survey-answers?hsLang=en). Japan's 7.4% "don't know," against 0.3% in each of the other three countries, probably carries some of that.

## How fast the gap is closing

Fast. In the previous year's survey (fiscal 2024, reported in the 2025 White Paper), only 49.7% of Japanese firms had a use policy, against 84.8% in the US. One year later Japan is up 19.2 points and the US is slightly down. The gap shrank from about 35 points to about 14.

Individual use moved too. In the companion consumer survey (January to February 2026; 1,236 people in Japan and 624 in each other country, ages 15 to 69), 58.8% of Japanese respondents had used a generative AI service. The US and Germany were both at 75.6%, China at 93.6%. Japan's figure looks like it more than doubled from 26.7% a year earlier, but this round added 15 to 19 year olds, so the two aren't directly comparable. For ages 20 and up, Japan is at 52.2%.

Among companies that know their own policy, actual use of generative AI in at least one task is 86.4% in Japan, 90.9% in the US, 91.6% in Germany and 98.1% in China. That's close. The difference is in how deep the use goes. In Japan the most common use is drafting meeting minutes and emails, at roughly 70% of firms, according to the White Paper.

## Why can't Japanese firms decide?

The usual answer is that Japanese companies worry more about risk. The data only partly supports that.

Japanese firms do lead on one concern. 46.4% cite security risks such as leaks of internal information, the highest of the four countries. But only 35.3% worry about output accuracy, compared with 44.3% in the US, 41.7% in Germany and 52.1% in China. Japan is also lowest on worries about biased output and privacy. Where Japan really stands out is "don't know": 16.5%, against 1 to 2% elsewhere.

Firms elsewhere worry more about accuracy and privacy, and they still set a policy. My reading is that in Japan the issue is ownership. Nobody's job is to make the call. The survey backs the first half of that: 27.0% of Japanese firms report no organized effort on generative AI at all, against 1.4% in the US, 4.9% in Germany and 2.6% in China. Among Japanese small and mid-sized firms it's 45.6%.

The second half is my interpretation. The survey didn't measure it. In many Japanese organizations, whoever makes an unprecedented call ends up owning its consequences. So when a topic has no clear owner, it waits. People often put this as "nobody wants to go first," and that matches what I see. Still, treat it as a hypothesis.

## What this means if you're selling AI into Japan

Your first job is to get past "we haven't decided." Find the person inside the undecided company who has taken on the job of deciding, and give them what they need to get it approved.

That person usually runs a pilot in a DX team or in corporate IT. They don't need another feature demo. They need one page that says where input data is stored, whether it's used for training, and who can see the logs. The most common risk measure among Japanese firms is a company-wide guideline (41.1%), so offering a draft internal usage guideline alongside the product tends to move things along. IT security and legal are where the worry is highest, so bring them in early. When they're comfortable, the approval path opens up.

For small and mid-sized firms, the numbers argue against direct sales. 31.5% have no clear policy and 45.6% have no organized effort. Visiting them one by one takes more time than it pays back. The realistic route is through the system integrators, resellers and local IT vendors they already rely on. A reseller agreement alone won't produce sales, though, for reasons I wrote about in [why your distributor in Japan isn't selling](https://nobusato.jp/en/blog/why-your-distributor-isnt-selling?hsLang=en).

And plan for the second sale. Japanese firms adopted cloud at high rates and then mostly used it for file storage, as I described in [what Japan's 80% cloud use is actually for](https://nobusato.jp/en/blog/eighty-percent-use-cloud-what-for?hsLang=en). Generative AI could stall at minutes and emails the same way. A plan for the second department and the third use case is worth as much as the first contract.

This survey only covers firms that have started digitalizing, so treat its levels with care. The 20.2% with no clear policy still says a lot about where the market is stuck. If your pipeline is full of Japanese firms that are still considering, you are looking at that 20.2% in your own data. Budget sales time for helping them write a policy before the demo.

Sources

- [MIC, 2026 White Paper on Information and Communications (Japanese): generative AI use policy by country, Fig. I-2-1-1 (Japan 68.9%, US 82.8%, Germany 82.2%, China 95.4%; includes fiscal 2024 reference values)](https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r08/html/nd121110.html)
- [MIC, 2026 White Paper: organized efforts on generative AI (none: Japan 27.0%)](https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r08/html/nd121120.html)
- [MIC, 2026 White Paper: business use of generative AI (at least one task: Japan 86.4%)](https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r08/html/nd121130.html)
- [MIC, 2026 White Paper: perceived risks and countermeasures (security risk: Japan 46.4%; company-wide guideline: 41.1%)](https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r08/html/nd121140.html)
- [MIC, 2026 White Paper: individual use of generative AI by country (Japan 58.8%, US 75.6%, Germany 75.6%, China 93.6%)](https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r08/html/nd111120.html)
- [Kyodo News via Kumamoto Nichinichi Shimbun, July 24, 2026 (Japanese): ages 20 and up, 52.2%](https://kumanichi.com/articles/2017202)
- [MIC, 2026 White Paper, appendix notes: survey outline (online panel, January to February 2026, 1,442 firms and 3,108 individuals)](https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r08/pdf/01fuchuu.pdf)
- [MIC, 2025 White Paper: AI use in companies (fiscal 2024 survey, Japan 49.7%)](https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r07/html/nd112220.html)

---

Shorter notes that don't make it into a full article go on LinkedIn. [Follow me there](https://www.linkedin.com/in/realnobusato/).

## More in this column

- ### [Japan's no-successor bankruptcies hit 264 in six months](https://nobusato.jp/en/blog/no-successor-bankruptcies-japan?hsLang=en)
  
  Tokyo Shoko Research counted 264 no-successor bankruptcies in Japan in H1 2026, a record. In 85.6% the president died ...
- ### [Japan's Labor-Shortage Bankruptcies Hit 427: Who Fell](https://nobusato.jp/en/blog/labor-shortage-bankruptcies-record?hsLang=en)
  
  Teikoku Databank counted 427 labor-shortage bankruptcies in Japan in 2025, a third straight record. 77% had under ten ...
- ### [Japan's Outward FDI Is 6.6 Times Its Inward FDI](https://nobusato.jp/en/blog/japan-outward-vs-inward-fdi?hsLang=en)
  
  At end-2025 Japan held ¥384.5 trillion of direct investment abroad and foreign firms held ¥58.6 trillion in Japan. What ...

[Column guide: Japan by the Numbers](https://nobusato.jp/en/blog/guide-japan-by-the-numbers?hsLang=en) · [Glossary](https://nobusato.jp/en/blog/glossary?hsLang=en)

[← All articles](https://nobusato.jp/en/blog?hsLang=en)

## Writing about launching in Japan, in three languages.

[Profile](https://nobusato.jp/en/profile?hsLang=en) [Get in touch](https://nobusato.jp/en/contact?hsLang=en)

Nobu Sato

Writing about launching in Japan, in three languages.

[info@nobusato.jp](mailto:info@nobusato.jp) [LinkedIn](https://www.linkedin.com/in/realnobusato/) [X](https://x.com/nobusato5888)

Column guides[Launching in Japan](https://nobusato.jp/en/blog/guide-launching-in-japan?hsLang=en)[Organizations & Hiring](https://nobusato.jp/en/blog/guide-organizations-and-hiring?hsLang=en)[Japan by the Numbers](https://nobusato.jp/en/blog/guide-japan-by-the-numbers?hsLang=en)[Career](https://nobusato.jp/en/blog/guide-career?hsLang=en)[Glossary](https://nobusato.jp/en/blog/glossary?hsLang=en)

© 2026 Nobu Sato. All rights reserved.

```json
{
  "@context" : "https://schema.org",
  "@type" : "BreadcrumbList",
  "itemListElement" : [ {
    "@type" : "ListItem",
    "item" : "https://nobusato.jp/en",
    "name" : "Home",
    "position" : 1
  }, {
    "@type" : "ListItem",
    "item" : "https://nobusato.jp/en/blog",
    "name" : "Articles",
    "position" : 2
  }, {
    "@type" : "ListItem",
    "item" : "https://nobusato.jp/en/blog/japan-generative-ai-adoption-gap",
    "name" : "Japan Trails on Generative AI Policy, Mostly by Not Deciding",
    "position" : 3
  } ]
}
```

```json
{
  "@context" : "https://schema.org",
  "@id" : "https://nobusato.jp/en/blog/japan-generative-ai-adoption-gap",
  "@type" : "WebPage",
  "author" : {
    "@id" : "https://nobusato.jp/#person"
  },
  "inLanguage" : "en",
  "isPartOf" : {
    "@id" : "https://nobusato.jp/en/blog",
    "@type" : "Blog",
    "name" : "Articles"
  },
  "name" : "Japan Trails on Generative AI Policy, Mostly by Not Deciding",
  "url" : "https://nobusato.jp/en/blog/japan-generative-ai-adoption-gap"
}
```

```json
{
  "@context" : "https://schema.org",
  "@id" : "https://nobusato.jp/#person",
  "@type" : "Person",
  "name" : "Nobu Sato",
  "sameAs" : [ "https://www.linkedin.com/in/realnobusato/", "https://x.com/nobusato5888" ],
  "url" : "https://nobusato.jp/en/profile"
}
```

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Nobu Sato",
    "url" : "https://nobusato.jp/en/blog/author/nobu-sato"
  },
  "dateModified" : "2026-10-24T14:18:24.677Z",
  "datePublished" : "2026-10-24T12:32:03.000Z",
  "headline" : "Japan Trails on Generative AI Policy, Mostly by Not Deciding",
  "image" : [ "https://nobusato.jp/hubfs/japan-generative-ai-adoption-gap-en.png" ],
  "mainEntityOfPage" : {
    "@id" : "https://nobusato.jp/en/blog/japan-generative-ai-adoption-gap",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "url" : "https://nobusato.jp/hubfs/nobusatojp-1.jpg"
    },
    "name" : "nobusato.jp"
  }
}
```