Curated AI News from Japan · 2026-10-13

What's happening in Japan today

6 of 28 Japanese-language candidates. Sources are a hand-picked whitelist, the ordering comes from model scoring, and the bar is whether it is useful to someone building AI in Japan — each entry says why we picked it. Most of this exists only in Japanese and never surfaces in English-language feeds.

2026-10-13 2026-10-12 2026-10-11 2026-10-10 2026-10-09 2026-10-08 2026-10-07 2026-10-06 2026-10-05 2026-10-04 2026-10-03 2026-10-02 2026-10-01 2026-09-30

01 One in Three Companies Exceeds Its Token Budget: How to Cut Costs by up to 44%【Accenture Survey】

3社に1社がトークン予算超過 コスト増を最大44%抑える方法とは【アクセンチュア調査】

MediaITmedia AI+

Accenture found that about 80% of token spending lacks clear business outcomes. With one in three companies over budget, leading firms manage tokens as a business resource.

Why choose itA Japanese enterprise’s generative-AI spending exceeded its budget, offering practical lessons in cost controls and token governance.

Read original →

02 Transcribe Web Meetings Without a Bot: ChatSense Desktop Feature Coming Soon

Web会議を、Botなしで文字起こし。ChatSenseのデスクトップ機能で対応予定

CompanyPR TIMES

Knowledge Sense plans to let its enterprise AI agent ChatSense transcribe meetings through a Windows desktop app without a bot joining the meeting.

Why choose itA Japanese company has introduced bot-free meeting transcription, making this a useful example for evaluating team workflows and adoption.

Read original →

03 supa.ai Launches Same-Day Penetration Testing to Reproduce Cyberattacks on the Latest AI Models

supa.ai、最新AIモデルのサイバー攻撃を再現するペネトレーションテストを最短即日実施を開始

MediaASCII.jp

supa.ai launches penetration tests that reproduce cyberattacks on the latest AI models, with completion possible on the same day.

Why choose itMedia reports on AI-model penetration-testing services provide a useful reference for assessing model security.

Read original →

04 SherLOCK Expands AI Red-Teaming Tests for AI Agents Amid a Wave of Cyberattacks

SherLOCK、相次ぐサイバー攻撃を背景にAIエージェント向けAIレッドチーミングテスト提供を本格化

CompanyPR TIMES

SherLOCK is expanding its AI red-teaming testing services for AI agents amid a series of cyberattacks.

Why choose itA local vendor’s red-team testing service for AI agents can help teams evaluate security before putting agents into production.

Read original →

05 I Asked AI to Create Horyu-ji’s Five-Story Pagoda — An Astonishing Result After Repeated Corrections

AIに「法隆寺の五重塔」を作らせてみた 修正を重ねてたどり着いた驚きの完成度

MediaASCII.jp

This article covers turning Horyu-ji’s five-story pagoda into a 3D model with Dots and running it in UE5. Revisions based on expert feedback and papers highlighted AI’s potential, structural understanding, and the need for expert supervision.

Why choose itThis case study of AI modeling for Japanese cultural architecture highlights a workflow that combines iterative revision with expert review.

Read original →

06 OrcaRouter Begins Offering Frontier Cybersecurity Model “OrcaCyber Zero 1.5”

OrcaRouter、サイバーセキュリティ向けフロンティアモデル「OrcaCyber Zero 1.5」の提供を開始

CompanyPR TIMES

FlashLabs announced that OrcaRouter will exclusively offer Continuum AI’s cybersecurity model “OrcaCyber Zero 1.5” in the Japanese market.

Why choose itAn exclusive offering from Japan provides a gateway for cybersecurity-focused models, giving teams an opportunity to assess integration and trial use.

Read original →