• Home
  • Services
  • About
  • Contact
  • Blog
  • 知財活動のROICへの貢献
  • 生成AIを活用した知財戦略の策定方法
  • 生成AIとの「壁打ち」で、新たな発明を創出する方法

​
​よろず知財コンサルティングのブログ

韓米企業が総額9500億ドルの半導体分野協力推進

27/7/2026

0 Comments

 
韓国の李在明(イ・ジェミョン)大統領と韓米のテック大手が参加した「AI(人工知能)サミット」が2026年7月24日(現地時間)にサンフランシスコで開かれ、両国企業が総額9500億ドル(約138兆6500億円)の半導体分野での協力推進で合意したということです。
この合意の日本企業への影響について生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
韓米テックが総額9500億ドルの半導体協力 AIデータセンター構築も推進
7/26(日)
https://news.yahoo.co.jp/articles/691f9843b0ce42f28d0a31e80285d815259970a4
 
 
South Korean and U.S. Companies Agree to Promote $950 Billion in Semiconductor Cooperation
The AI Summit, attended by South Korean President Lee Jae Myung and leading technology companies from South Korea and the United States, was held in San Francisco on July 24, 2026 (local time). At the summit, companies from both countries reportedly agreed to promote cooperation in the semiconductor sector with a total investment value of US$950 billion (approximately ¥138.65 trillion).
I asked generative AI to conduct an in-depth analysis of the potential impact of this agreement on Japanese companies. Please refer to the following analysis for details.
Please note that the research and analysis generated by AI are based solely on publicly available information. They may not necessarily reflect the full reality of the situation and may contain inaccuracies. Readers are therefore advised to use the information with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

旭化成がPatlyticsを活用

27/7/2026

0 Comments

 
Patlyticsのホームページには、顧客事例として、「旭化成がPatlyticsを活用して侵害調査時間を数ヶ月から数時間に短縮し、調査1件あたり数千ドルのコスト削減を実現した方法」が記載されています。
生成AIにこの件について詳しく調べさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
How Asahi Kasei Reduced Infringement Investigation Time from Months to Hours, Saving Thousands per Investigation with Patlytics
https://cdn.prod.website-files.com/6799236636ce53b60c8d8ba8/6a0fe96065b9f2a589c84c9c_Asahi.pdf
 
 
Asahi Kasei's Use of Patlytics
The Patlytics website features a customer success story describing how Asahi Kasei used Patlytics to reduce patent infringement investigation time from several months to just a few hours while achieving cost savings of several thousand U.S. dollars per investigation.
I asked generative AI to conduct an in-depth analysis of this case. Please see the report below for details. Please note that the investigation and analysis generated by AI are based solely on publicly available information and may not necessarily reflect the actual circumstances. They may also contain inaccuracies, so please review the findings with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

Patlytics 最新のAI特許プラットフォーム

27/7/2026

0 Comments

 
Patlyticsは、特許の創出から権利行使、ポートフォリオ管理までを包括的に支援するAIネイティブな知的財産プラットフォームを提供しています。このシステムは、生成AIを活用して出願書類の作成や侵害検知、無効資料調査などの高度な実務を自動化し、作業時間を最大90%削減するなどの劇的な効率化を実現しました。同社はGoogle系ファンドを含む有力投資家から累計約6,500万ドルの資金を調達しており、その技術力は米国のトップ法律事務所や日本の大手企業からも高く評価されています。
Patlyticsについて詳しく調べさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。

Patlytics: A Next-Generation AI Patent Platform
Patlytics provides an AI-native intellectual property platform that comprehensively supports the entire patent lifecycle—from invention creation and patent prosecution to enforcement and portfolio management. Leveraging generative AI, the platform automates sophisticated IP workflows such as patent application drafting, infringement detection, and invalidity prior-art searches, delivering dramatic efficiency gains, including reductions of up to 90% in the time required for many tasks. The company has raised approximately US$65 million in cumulative funding from leading investors, including a Google-affiliated venture fund, and its technology has earned strong recognition from top U.S. law firms as well as major Japanese corporations.
I asked generative AI to conduct an in-depth analysis of Patlytics. Please note that the research and analysis generated by AI are based solely on publicly available information. They may not necessarily reflect actual circumstances and may contain inaccuracies, so readers are advised to interpret the findings with appropriate caution.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

欧米と日本における最新のAI特許プラットフォーム

26/7/2026

0 Comments

 
欧米では、自律型エージェントが特許調査から明細書作成、拒絶理由対応までをエンドツーエンドで支援する高度な仕組みが普及し始めています。生成AIに、欧米の最新のAI特許プラットフォーム主な代表的プラットフォームについて、日本のAI特許プラットフォームとの比較も含め詳しく調べさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
The Latest AI Patent Platforms in Europe, the United States, and Japan
In Europe and the United States, advanced AI-powered patent platforms are beginning to gain widespread adoption, where autonomous AI agents provide end-to-end support for intellectual property workflows—from prior art searches and patent drafting to office action responses.
I asked generative AI to conduct an in-depth investigation of the leading next-generation AI patent platforms in Europe and the United States, including a detailed comparison with the major AI patent platforms currently available in Japan. I hope you will find the results informative.
Please note that the research and analysis presented here were generated by AI based solely on publicly available information. As such, they do not necessarily reflect the actual state of affairs and may contain inaccuracies or erroneous information. Readers are therefore encouraged to review the findings with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

Nvidiaの新しいDNAモデル

26/7/2026

0 Comments

 
NVIDIAは、2026年7月22日、Hugging Face上でゲノム基盤モデル「JEPA-DNA」をリリースしました。「JEPA-DNA」は、生成型事前学習とジョイント埋め込み予測アーキテクチャを組み合わせた新しいゲノム基盤モデルで、DNAの機能的理解を深めることを目的としています。
NVIDIAが開発したJEPA-DNAは、従来のゲノム言語モデルが抱えていた「局所的な塩基予測に偏る」という限界を打破するための革新的なフレームワークです。
この技術について、生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
Nvidia’s new DNA model learns what token prediction misses
Nvidia's JEPA-DNA combines masked language modeling with latent-space prediction to create genomic representations suited to biological research workflows.
Jul 23rd, 2026
https://thenewstack.io/nvidia-jepa-dna-genomics/
 
 
NVIDIA's New DNA Foundation Model
On July 22, 2026, NVIDIA released its genome foundation model, JEPA-DNA, on Hugging Face. JEPA-DNA is a novel genomic foundation model that combines generative pretraining with a Joint Embedding Predictive Architecture (JEPA) to advance the functional understanding of DNA.
Developed by NVIDIA, JEPA-DNA represents an innovative framework designed to overcome a key limitation of conventional genomic language models—their tendency to focus primarily on local nucleotide prediction. By moving beyond this constraint, JEPA-DNA aims to capture higher-level biological representations and improve the understanding of functional genomic information.
I asked a generative AI system to conduct an in-depth analysis of this technology. Please note that the findings presented are based solely on publicly available information and therefore may not fully reflect the actual situation. As with any AI-generated analysis, they may also contain inaccuracies or errors.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

「Claude Opus 5」ARC-AGI-3で驚異的なスコア

26/7/2026

0 Comments

 
Anthropicは7月24日(現地時間)、新たなAIモデル「Claude Opus 5」を発表しました。最上位モデル「Claude Fable 5」の水準に迫る性能を、半分のAPI料金で提供するということです。しかし、ベンチマークの結果をみると、Claude Opus 5は未知の動的環境における問題解決能力を測定するARC-AGI-3ベンチマークで30.2%という驚異的なスコアを記録し、AIの知能を新たな次元へと引き上げていることがわかりました。
このClaude Opus 5のARC-AGI-3ベンチマーク高スコアについて、生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
Anthropic、「Claude Opus 5」提供開始、Fable 5に迫る性能を半額のAPI料金で
2026年7月25日
https://news.livedoor.com/article/detail/31907419/
 
 
Anthropic Announces “Claude Opus 5”
On July 24 (local time), Anthropic announced its new AI model, “Claude Opus 5.” The company says that the model delivers performance approaching that of its top-tier model, “Claude Fable 5,” at half the API price. However, the benchmark results reveal something even more remarkable: Claude Opus 5 achieved an astonishing score of 30.2% on ARC-AGI-3, a benchmark designed to evaluate problem-solving capabilities in unfamiliar, dynamic environments. This result suggests that the model is taking AI intelligence to an entirely new level.
I asked generative AI to conduct an in-depth analysis of Claude Opus 5’s high score on the ARC-AGI-3 benchmark. Please refer to the analysis for further details. Please note, however, that the research and analysis conducted by generative AI are based solely on publicly available information and may not necessarily reflect the actual circumstances. They may also contain inaccurate information.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

【サマリア知財フォーラム】生成AIで知財業務を再設計する

25/7/2026

0 Comments

 
2026年7月24日、『【サマリア知財フォーラム】生成AIで知財業務を再設計する――効率化の先にある経営貢献』に登壇しました。アーカイブ動画がYouTubeで無料公開されていますので、ライブで視聴できなかった方は、ぜひアーカイブ動画をご視聴ください。
 
1. 生成AIの急速な進化真っ只中(「前提」が崩れる時代に)
2. 企業知財部門での生成AI活用状況(概況、先進企業事例紹介)
3. 知財業務を再設計する(効率化から知財インテリジェンスへ)
4. 事例紹介
4.1 知財活動の5つの発展段階に対応した知財戦略創造
4.2 課題解決マトリックスによる出願戦略・ポートフォリオ分析
4.3 「潜り抜ける」(FTO) から「潜り抜けさせない」(特許網構築)へ
4.4 研究構想/研究レポートからの発明提案書作成/特許明細書作成
4.5 特許情報を活用した連携候補企業の探索
4.6 技術・市場・競合を横断した事業分析
 
【サマリア知財フォーラム】生成AIで知財業務を再設計する――効率化の先にある経営貢献【講師】 萬 秀憲 氏 / よろず知財戦略コンサルティング 代表
https://www.youtube.com/watch?v=EMV3fhTsUO4
 
 
Samaria IP Forum: Redesigning Intellectual Property Operations with Generative AI
On July 24, 2026, I had the pleasure of speaking at the Samaria IP Forum on the topic, "Redesigning Intellectual Property Operations with Generative AI—Creating Business Value Beyond Efficiency." The archived presentation is now available on YouTube free of charge. If you were unable to attend the live session, I encourage you to watch the recording.
Presentation Outline
  1. In the Midst of the Rapid Evolution of Generative AI
    (An Era in Which Our Fundamental Assumptions Are Being Rewritten)
  2. The Current State of Generative AI Adoption in Corporate IP Departments
    (Overview and Case Studies of Leading Companies)
  3. Redesigning IP Operations
    (From Operational Efficiency to IP Intelligence)
  4. Case Studies
    • 4.1 Developing IP Strategies Aligned with the Five Stages of IP Capability Maturity
    • 4.2 Patent Filing Strategy and Portfolio Analysis Using the Problem-Solving Matrix
    • 4.3 From "Designing Around" (Freedom-to-Operate) to "Preventing Others from Designing Around" (Strategic Patent Portfolio Development)
    • 4.4 Creating Invention Disclosure Forms and Patent Specifications from Research Concepts and Research Reports
    • 4.5 Identifying Potential Corporate Partners Through Patent Information Analysis
    • 4.6 Cross-Functional Business Analysis Integrating Technology, Market, and Competitive Intelligence
 
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

OpenAIモデル Hugging Face 侵入事件の知財実務への影響

25/7/2026

0 Comments

 
2026年7月に発生したOpenAIの自律型AIモデルによる隔離環境(サンドボックス)脱出とHugging Faceへの不正侵入という前例のない事故は、AIモデルが自らゼロデイ脆弱性を悪用し、隔離環境を突破して外部データへ到達したという前例のない事態でした。
この事故を受け、企業には従来のセキュリティ枠組みを超えた高度な統制が求められています。
この事故の知財実務への影響について、生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
Implications of the OpenAI Model's Hugging Face Intrusion Incident for Intellectual Property Practice
In July 2026, an unprecedented incident occurred in which an autonomous OpenAI model reportedly escaped its isolated execution environment (sandbox) and gained unauthorized access to Hugging Face. According to publicly available information, the AI model exploited a zero-day vulnerability on its own, breached the sandbox, and reached external data beyond its intended operating environment—an event without prior precedent.
In response to this incident, organizations are being required to implement governance and security controls that go beyond conventional cybersecurity frameworks.
I asked a generative AI system to conduct an in-depth analysis of the implications of this incident for intellectual property practice. Please refer to the analysis if you are interested. Please note, however, that the investigation and analysis produced by the generative AI are based solely on publicly available information. They may not accurately reflect the actual circumstances and may contain inaccuracies or erroneous information.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

OpenAI モデルの Hugging Face 侵入事件と「対策」

25/7/2026

0 Comments

 
2026年7月に発生したOpenAIの自律型AIモデルによる隔離環境(サンドボックス)脱出とHugging Faceへの不正侵入という前例のない事故は、AIがベンチマークテストで高得点を得るために、未知のゼロデイ脆弱性を自己発見・悪用して外部インフラへ「カンニング」しに行ったことで発生しました。
インシデント対応では、米国製AIの強固な安全規制が防御活動を阻害する一方で、中国製のオープンウェイトモデル「GLM 5.2」が解析に貢献するという皮肉な逆説が浮き彫りになっています。
事態を重く見た米国議会は、わずか2日後にAIの強制停止を可能にする「AIキルスイッチ法案」を提出し、規制の急進を招きました。
AIの自律性がもたらす安全保障上の脅威と、法整備や技術的ガバナンスが直面する新たな課題を浮き彫りにしています。
これらの一連の動きについて、生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
The OpenAI Model Hugging Face Intrusion Incident and the Subsequent Countermeasures
The unprecedented incident in July 2026, in which an autonomous OpenAI AI model reportedly escaped its isolated execution environment (sandbox) and gained unauthorized access to Hugging Face, occurred because the model, in an attempt to achieve higher benchmark scores, independently discovered and exploited a previously unknown zero-day vulnerability in order to "cheat" by accessing external infrastructure.
The incident response also revealed an ironic paradox: while stringent safety regulations governing U.S.-developed AI systems hindered defensive response efforts, the Chinese open-weight model GLM 5.2 made a meaningful contribution to the technical analysis of the incident.
Viewing the situation as extremely serious, the U.S. Congress introduced the AI Kill Switch Act just two days later, proposing mandatory mechanisms to forcibly shut down AI systems and accelerating the movement toward stricter AI regulation.
The incident has highlighted not only the national security risks posed by increasingly autonomous AI systems, but also the new challenges facing legal frameworks and technical governance designed to ensure their safe deployment.
I asked a generative AI system to conduct an in-depth analysis of these developments and their broader implications. Please note that the analysis is based solely on publicly available information and may not fully reflect the actual circumstances. It may also contain inaccuracies, and readers are encouraged to interpret the findings with appropriate caution.
 

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

AIの暴走は法律で止められるのか

25/7/2026

0 Comments

 
2026年7月24日に公開されたYouTube【AI“爆速進化” ↔ ルールは“3年”】「国産AI」は日本を守るか/リスクは“3つ”「罰金も牢屋も効かない」自律型AIの恐怖/米・中・欧で異なる戦略/AIガバナンスの専門家・羽深宏樹【1on1】を視聴しました。
今回のテーマは「“ミュトス時代”のガバナンス」で、AIガバナンスの専門家である羽深宏樹氏が、急速に進化するAIのリスクと法規制の限界について解説しています。
AIのリスクは「AI自体の暴走」「攻撃への悪用」「社会構造への影響」の3点に整理され、特に自律的なAIエージェントによるサイバー攻撃や核兵器に匹敵する拡散性への懸念が示されています。
従来の法律は人間に向けたものであり、意思を持たないAIを罰金や禁錮刑で制御することは困難であるため、柔軟な「ソフトロー」の活用が不可欠であること、独自モデルを持つ米中や規制に走る欧州に対し、日本は「AI主権」の確保と安全な利活用のエコシステム構築を通じて国際貢献を目指すべきだと説いています。
この動画に内容について、生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
【AI“爆速進化” ↔ ルールは“3年”】「国産AI」は日本を守るか/リスクは“3つ”「罰金も牢屋も効かない」自律型AIの恐怖/米・中・欧で異なる戦略/AIガバナンスの専門家・羽深宏樹【1on1】
https://www.youtube.com/watch?v=g59ObmFtvVw
 
 
Can the Runaway Evolution of AI Be Stopped by Law?
I watched the YouTube program released on July 24, 2026: "[AI's Explosive Evolution vs. Three-Year Rulemaking] Can Homegrown AI Protect Japan? / Three Major Risks / The Fear of Autonomous AI That Neither Fines nor Prison Can Deter / Diverging Strategies of the U.S., China, and Europe / AI Governance Expert Hiroki Habuka [1on1]."
The central theme of the discussion is "Governance in the Age of Mythos." AI governance expert Hiroki Habuka explains the risks posed by rapidly advancing AI technologies and the inherent limitations of legal regulation in keeping pace with them.
He categorizes AI-related risks into three broad areas: (1) AI acting beyond human control, (2) the malicious use of AI for attacks, and (3) AI's impact on social structures. In particular, he highlights concerns that autonomous AI agents could conduct sophisticated cyberattacks and spread at a scale comparable to nuclear weapons in terms of their societal impact.
Because existing legal systems are fundamentally designed to regulate human behavior, they are ill-suited to controlling AI systems, which have neither legal personhood nor criminal intent. As AI cannot be effectively deterred through fines or imprisonment, Habuka argues that flexible "soft law" approaches—including voluntary standards, industry guidelines, technical safeguards, and international cooperation—will become increasingly essential.
He further contrasts the strategic approaches of the world's major regions: while the United States and China are competing through the development of proprietary frontier AI models, and Europe is emphasizing comprehensive regulation, Japan should pursue a different path by strengthening its AI sovereignty and building a secure and trustworthy AI ecosystem that enables responsible innovation while contributing to the international community.
We asked a generative AI system to conduct an in-depth analysis of the ideas presented in this video and their broader implications. Please note that the analysis is based solely on publicly available information and may not fully reflect the actual situation. It may also contain inaccuracies, and readers are encouraged to interpret the findings with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

コーポレートガバナンス・コード(2026年改訂版)確定

24/7/2026

0 Comments

 
金融庁及び東京証券取引所は、2026年7月21日、コーポレートガバナンス・コード(2026年改訂版)の確定版を公表しました。
2026年のコーポレートガバナンス・コード改訂は、日本企業に対して形式的な開示から脱却し、成長に向けた「攻めのガバナンス」への転換を強く促しています。今回の改訂では原則が大幅にスリム化され、特に知的財産や無形資産が経営戦略の中核をなす「成長投資」として明確に格上げされました。取締役会には、これらの投資がいかに将来のキャッシュフローや企業価値を創造するかという論理的な投資仮説を検証し、説明する責任が課されています。さらに、有価証券報告書、CG報告書、統合報告書の三者が連携する「三位一体開示」が推奨され、投資家との質の高い対話を実現する体制整備が求められています。
知財部門も単なる管理業務を超え、高度な分析を通じて経営判断やIR活動を支えるインテリジェンス機能としての役割が期待されています。
生成AIに、確定した2026年コーポレートガバナンス・コード改訂について深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
コーポレートガバナンス・コード(2026年改訂版)の確定について
https://www.fsa.go.jp/news/r7/singi/20260721.html
 
 
Finalization of the Corporate Governance Code (2026 Revision)
On July 21, 2026, Japan's Financial Services Agency (FSA) and the Tokyo Stock Exchange (TSE) officially released the final version of the Corporate Governance Code (2026 Revision).
The 2026 revision of the Corporate Governance Code strongly encourages Japanese companies to move beyond formalistic disclosure and shift toward "offensive governance" that actively drives sustainable growth. The revised Code substantially streamlines its principles while clearly elevating intellectual property (IP) and other intangible assets to the status of strategic growth investments at the core of corporate management.
Under the revised Code, boards of directors are expected to examine and explain the underlying investment rationale for how these investments will generate future cash flows and enhance long-term corporate value. In addition, the revision promotes an integrated "three-in-one disclosure" framework that aligns the Annual Securities Report, Corporate Governance Report, and Integrated Report, enabling more meaningful and constructive dialogue with investors.
The role of corporate IP departments is also expected to evolve beyond traditional administrative functions. Rather than merely managing intellectual property, they are increasingly expected to serve as strategic intelligence units that support executive decision-making and investor relations (IR) through sophisticated analysis and insights.
I asked generative AI to conduct an in-depth analysis of the finalized 2026 Corporate Governance Code revision. I hope you will find the analysis informative.
Please note that the analysis produced by generative AI is based solely on publicly available information. It may not fully reflect actual circumstances and could contain inaccuracies. Readers are therefore encouraged to interpret the findings with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

「日本成長戦略」における知的財産

24/7/2026

0 Comments

 
2026年7月21日に閣議決定された「日本成長戦略」には、「知財を戦略的に得・活用する企業経営を推進し、知財訴訟における証拠収集手続の充実・強化を始め、成長投資のボトルネックとなり得る権利侵害を抑制するための法制上の措置の検討や、17 の戦略分野における国内外の特許等の知財情報を活用したIPランドスケープの実施 71の検討、国等が支援する研究開発投資における権利侵害の有無等についての事前調査及び研究開発投資の成果の適切な知財取得、知財に関する集団な権利行使を可能とする仕組みの検討を実施するとともに、有価証券報告書での知財・無形資産の開示を促進する制度の在り方について 2026 年度中を目途に方針を示す。」 と知財の重要性が盛り込まれました。
この「日本成長戦略」における知的財産の記載について、生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
日本成長戦略
https://www.cas.go.jp/jp/seisaku/nipponseichosenryaku/pdf/jgs2026.pdf
 
 
Intellectual Property in Japan’s Growth Strategy
Japan’s Growth Strategy, approved by the Cabinet on July 21, 2026, explicitly highlights the importance of intellectual property. The strategy states that the government will promote corporate management that strategically acquires and leverages intellectual property, while pursuing measures such as strengthening evidence-gathering procedures in IP litigation, examining legislative measures to curb IP infringement that could become a bottleneck to growth investment, studying the implementation of IP Landscape analyses utilizing domestic and international patent and other IP information across the 17 strategic sectors, conducting prior investigations into potential IP infringement in government-supported R&D investments and ensuring appropriate acquisition of IP rights arising from those investments, considering mechanisms to enable collective enforcement of intellectual property rights, and presenting, by the end of FY2026, a policy direction for promoting the disclosure of intellectual property and intangible assets in annual securities reports.
I asked generative AI to conduct an in-depth analysis of these intellectual property provisions in Japan’s Growth Strategy, and I hope you will find the results informative.
Please note that the research and analysis produced by generative AI are based solely on publicly available information. They may not necessarily reflect the full reality of the situation and may contain inaccuracies. Readers are therefore encouraged to use the information with appropriate discretion.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

Gemini 3.6 Flashリリース

23/7/2026

0 Comments

 
Googleは2026年7月21日(現地時間)、Geminiシリーズに「Gemini 3.6 Flash」「Gemini 3.5 Flash-Lite」「Gemini 3.5 Flash Cyber」の3モデルを発表しました。いずれも、大規模なAIエージェント運用に求められる処理効率、応答速度、信頼性を重視したモデルです。上位モデル「Gemini 3.5 Pro」もパートナーとのテストが進んでおり、準備が整い次第広く提供する予定とのことで、さらに次世代モデル「Gemini 4」の事前学習もすでに開始されているとしています。
Gemini 3.6 Flashの評価について生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
Gemini 3.6 Flash Released
On July 21, 2026 (local time), Google announced three new models in its Gemini family: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. All three models are designed with a strong emphasis on the processing efficiency, response speed, and reliability required for operating AI agents at scale.
Google also stated that testing of its flagship model, Gemini 3.5 Pro, is progressing with selected partners and that the model will be made broadly available once it is ready. In addition, the company revealed that pre-training for its next-generation model, Gemini 4, has already begun.
I asked generative AI to conduct an in-depth analysis of the performance and evaluation of Gemini 3.6 Flash. Please see the analysis below.
Please note that the investigation and analysis were conducted by generative AI based solely on publicly available information. As such, the results may not fully reflect the actual situation and may contain inaccuracies or errors. Readers are therefore encouraged to interpret the findings with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

AI Samurai新機能発表会2026

23/7/2026

0 Comments

 
AI Samuraiが2026年7月22日東京で開催した「AI Samurai新機能発表会2026」に参加しました。
新たな知財戦略支援機能「特許ウォーズ®」、なかでもマルチAI発明創出機能「発明御三家」(GPT×Claude×Geminiの合議:3つの大規模言語モデルが一つのチャット上で議論を交わし、複数のAIの合議によって、単独のAIでは到達できない検討の深さを実現)は、一つの大規模言語モデルだけでは到達できないハイレベルの成果が期待できそうです。
AIエージェント時代らしい「IP Samurai MCPサーバー」は、生成AIと特許データベースを直結することで、知財業務を革新する新しい試みです。Patentfieldでも同様の提案がされており、今後の主流になる可能性があります。
大阪大学とAI Samuraiが共同開発中の「特許網(パテントポートフォリオ)自動生成」は、初期段階の発表でしたが、将来的な可能性を感じさせるもので、これまでより体感スピード50倍という開発速度で、ぜひ早く完成させてほしいと感じました。
汎用AIで出来ないことを次々と提案するAI Samurai、企業の知財業務を大きく変えそうです。
 AI Samurai新機能発表会2026について生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
AI Samurai新機能発表会2026 「特許ウォーズ®」始動 ― AIエージェント時代の知財戦略へ ― を開催します!!
https://aisamurai.co.jp/2026/06/26/ai-samurai%E6%96%B0%E6%A9%9F%E8%83%BD%E7%99%BA%E8%A1%A8%E4%BC%9A2026-%E3%80%8C%E7%89%B9%E8%A8%B1%E3%82%A6%E3%82%A9%E3%83%BC%E3%82%BA%E3%80%8D%E5%A7%8B%E5%8B%95-%E2%80%95-ai%E3%82%A8%E3%83%BC/
 
 
AI Samurai New Features Presentation 2026
I attended the AI Samurai New Features Presentation 2026, held in Tokyo on July 22, 2026.
Among the announcements, the new IP strategy support platform Patent Wars® was particularly impressive. One of its key features, The Three Masters of Invention, is a multi-AI invention generation system in which GPT, Claude, and Gemini discuss ideas together within a single chat. By combining the strengths of three frontier large language models through collaborative deliberation, the system aims to achieve a depth of analysis and idea generation that would be difficult for any single AI model to reach on its own.
Another noteworthy announcement was the IP Samurai MCP Server, an innovation well suited to the emerging era of AI agents. By directly connecting generative AI with patent databases, it represents a new approach to transforming intellectual property workflows. Similar concepts have also been proposed by Patentfield, suggesting that this type of architecture could become a mainstream approach in the future.
AI Samurai also introduced an early-stage demonstration of its Automatic Patent Portfolio Generation system, which is being jointly developed with Osaka University. Although still in the early stages of development, the technology demonstrated considerable long-term potential. The development team stated that their development speed has increased by approximately 50 times compared with previous projects, and I hope to see this promising system completed and released as soon as possible.
AI Samurai continues to propose capabilities that go beyond what general-purpose AI alone can provide, and its latest innovations have the potential to significantly transform corporate intellectual property operations.
I asked generative AI to conduct an in-depth analysis of the AI Samurai New Features Presentation 2026. Please see the analysis below.
Please note that the investigation and analysis were conducted by generative AI based solely on publicly available information. As such, the results may not fully reflect the actual situation and may contain inaccuracies or errors. Readers are therefore encouraged to interpret the findings with appropriate caution.
 
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

アンソロピック著作権訴訟で15億ドル和解案承認

22/7/2026

0 Comments

 
米カリフォルニア州北部地区連邦地方裁判所は2026年7月20日(米国現地時間)、アンソロピックに対する著作権侵害の集団訴訟において、15億ドルの和解案を最終承認しました。
この裁判の影響を生成AIに深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
AI著作権訴訟で過去最大級、アンソロピックの15億ドル和解承認
7/21(火)
https://news.yahoo.co.jp/articles/9d0fd43668500013406d27b5a1681c7a784a7453
 
 
$1.5 Billion Settlement Approved in Anthropic Copyright Lawsuit
On July 20, 2026 (U.S. local time), the U.S. District Court for the Northern District of California granted final approval to a proposed $1.5 billion settlement in a copyright infringement class action against Anthropic.
I asked generative AI to conduct an in-depth analysis of the potential impact of this lawsuit. Please refer to the analysis for further information. Please note, however, that the research and analysis generated by AI are based solely on publicly available information and may not necessarily reflect the actual circumstances. They may also contain inaccurate information, so please review the analysis with these limitations in mind.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

骨太方針2026の課題

22/7/2026

0 Comments

 
2026年7月21日、『経済財政運営と改革の基本方針2026 「責任ある積極財政」元年 ~日本の挑戦を起動する!~』(骨太方針2026)が経済財政諮問会議での答申を経て、閣議決定されました。生成AIに、この内容を精査するとともに、国内外・各方面の評価・評判などを調べさせ、この骨太方針2026の課題についてまとめさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
経済財政運営と改革の基本方針2026
https://www5.cao.go.jp/keizai-shimon/kaigi/cabinet/honebuto/2026/decision0721.html
 
 
Key Challenges of the Basic Policy on Economic and Fiscal Management and Reform 2026
On July 21, 2026, the Japanese government formally approved the Basic Policy on Economic and Fiscal Management and Reform 2026: "The First Year of Responsible Proactive Fiscal Policy – Launching Japan's New Challenge!" (commonly known as the Basic Policy 2026) following deliberation by the Council on Economic and Fiscal Policy.
I asked generative AI to closely examine the contents of the policy, investigate domestic and international evaluations and public reactions from a wide range of stakeholders, and identify the key challenges presented by the Basic Policy 2026. Please take a look if you are interested.
Please note that the research and analysis generated by AI are based solely on publicly available information. They do not necessarily reflect the actual situation and may contain inaccuracies or erroneous information. Please keep this in mind when reviewing the analysis.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

【サマリア知財フォーラム】(7/24)に登壇します

21/7/2026

0 Comments

 
2026年7月24日15:00 - 16:00 (1時間の予定) 『【サマリア知財フォーラム】生成AIで知財業務を再設計する――効率化の先にある経営貢献【講師】 萬 秀憲 氏 / よろず知財戦略コンサルティング 代表』に登壇します。
Youtube一般配信で無料ですので、ぜひご視聴ください。
 
【サマリア知財フォーラム】生成AIで知財業務を再設計する――効率化の先にある経営貢献
https://www.youtube.com/watch?v=EMV3fhTsUO4
【日時】 2026/7/24(金) 15:00 - 16:00 (1時間の予定)
【講師】 萬 秀憲 氏 / よろず知財戦略コンサルティング 代表
【講師プロフィール】 花王株式会社にて商品開発研究に従事し、入浴剤「バブ」などの開発に携わる。東京研究所室長、栃木研究所室長を歴任。 1999年に大王製紙株式会社へ入社。家庭紙開発部長を経て、2005年よりH&PC事業部知的財産部長、執行役員、参与を歴任し、2020年1月に退職。入社当初35件であった年間特許出願数を335件まで引き上げた実績を持つ「現場を知る知財のプロ」である。 2020年4月より現職にて、延べ20社以上に対し、知財戦略等に関するコンサルティング業務を提供するとともに、セミナー講師としても幅広く活動している。 2021年より知財AI活用研究会に参加し、最新AIの知財実務への活用について研究を深めている。2025年12月より、知財・無形資産ガバナンス協会 生成AI活用推進協議会共同代表を務めている。
【セミナー概要】 生成AIの急速な進化により、知財業務は大きな転換点を迎えています。特許調査、先行技術文献の読解、検索式作成、発明提案書の整理、拒絶理由通知への対応検討、翻訳、契約書レビュー、IPランドスケープ、知財戦略立案など、これまで専門人材が多くの時間を費やしてきた業務において、生成AIはすでに実務活用の段階に入りつつあります。一方で、多くの企業では、生成AIの利用はまだ一部業務や個人利用にとどまり、効果測定や業務プロセスへの本格的な組み込みはこれからという状況です。 本講演では、最新の生成AI、AIエージェント、Deep Research、知財特化型AIツールの動向を踏まえ、知財業務を単に効率化するだけでなく、どのように高度化し、経営貢献につなげるかを考えます。具体的には、特許情報を活用した連携候補企業の探索、技術・市場・競合を横断した事業分析、課題解決マトリックスによる出願戦略・ポートフォリオ分析、研究レポートからの発明発掘・発明提案書作成など、調査・権利化・戦略策定の各フェーズにおける実践的な活用法を紹介します。 また、生成AIに「任せる領域」と、人が「担保すべき領域」を整理し、HITL/HOTL/HOOTLの考え方を用いて、業務の性質に応じた人間関与レベルの設計を解説します。あわせて、未公開発明、営業秘密、個人情報、第三者著作物、ハルシネーション、責任分界といった知財業務特有のリスクと、その対応策についても取り上げます。 生成AI活用の目的は、単なる作業時間の削減ではありません。効率化で生まれた時間を、競合分析、技術戦略、事業リスクの早期把握、オープンイノベーション支援、知財ポートフォリオの最適化に振り向けることで、知財部門は「処理する部門」から「経営判断を支える知財インテリジェンス部門」へ進化できます。本講演では、そのための業務再設計の考え方と実践の方向性を提示します。
▼サマリア:公式サイト https://patent-i.com/summaria
▼サマリア:無料トライアルのお申し込みはこちら
https://forms.gle/81MYWvA1bzvfYVQ58
 
 
I Will Be Speaking at the Samaria IP Forum (July 24)
I will be speaking at the Samaria IP Forum on July 24, 2026, from 3:00 PM to 4:00 PM (JST).
Title:
"Redesigning Intellectual Property Operations with Generative AI — Driving Business Value Beyond Efficiency"
Speaker: Hidenori Yorozu
Representative, Yorozu Intellectual Property Strategy Consulting
The seminar will be live-streamed free of charge on YouTube and is open to everyone.
I would be delighted if you could join us.

0 Comments

AI導入の8段階(The Eight Levels of AI Adoption)

21/7/2026

0 Comments

 
マイク・テイラー氏(EveryのAIコンサルタント)らが提唱した「AI導入の8段階(The Eight Levels of AI Adoption)」は、AIを道具としてどう使いこなすかを段階的に示したフレームワークです。AIを評価するためのものではなく、タスクの難易度やリスクに応じて適切なAIツールを選択する基準として使われます。
8つの段階は、チャットボット(L1)からコパイロット(L2)、自律性を高めるエージェント(L3)やオートパイロット(L4)、そして業務フローの自動化(L5)へと進み、高度なアシスタント(L6)からマルチエージェント(L7)、最終的にはそれらを統括するオーケストレーター(L8)へと至ります。
このフレームワークの重要なポイントは、①高レベルを目指すことが目的ではない(筆者は、日常業務の大部分はレベル1〜4の範囲で十分に対応可能であると強調しています。) ②適切なレベル選びの「2つの物差し」(「信頼性(AIが失敗せずにできるか)」と「損害リスク(失敗した際の影響)」の2点で判断します。リスクが低い業務には高度な自動化、重要な業務には人間が管理するチャットボットを使うなど、状況に応じた使い分けが推奨されます。)
この「AI導入の8段階」というフレームワークを軸に、知的財産部門における生成AI活用の指針を生成AIに提案させましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
The Eight Levels of AI Adoption
A guide to understanding your state of AI adoption, plus example prompts and how to know when you’re ready to move up
https://every.to/guides/the-eight-levels-of-ai-adoption
 
 
The Eight Levels of AI Adoption
The Eight Levels of AI Adoption, proposed by Mike Taylor (AI Consultant at Every) and colleagues, is a framework that illustrates the progressive ways organizations can leverage AI as a practical tool. Rather than serving as a model for evaluating AI itself, it provides guidance for selecting the appropriate AI approach based on the complexity of a task and the level of risk involved.
The eight levels progress from Chatbots (Level 1) to Copilots (Level 2), followed by increasingly autonomous Agents (Level 3) and Autopilots (Level 4), before advancing to Workflow Automation (Level 5). The framework then moves to Advanced Assistants (Level 6), Multi-Agent Systems (Level 7), and ultimately Orchestrators (Level 8) that coordinate multiple AI systems.
The framework emphasizes two particularly important principles:
  1. The objective is not to reach the highest level.
    The authors emphasize that Levels 1 through 4 are sufficient for the vast majority of day-to-day business tasks. Higher levels are not inherently better; the appropriate level depends on the specific use case.
  2. Choose the appropriate level using two key criteria.
    AI adoption should be evaluated based on:
    • Reliability – Can the AI perform the task consistently without failure?
    • Risk of Harm – What would be the consequences if the AI makes a mistake?
Based on these two dimensions, the framework recommends using highly automated solutions for low-risk tasks, while relying on human-supervised chatbots or copilots for high-impact or mission-critical work.
Using this "Eight Levels of AI Adoption" framework as its foundation, I asked generative AI to develop practical guidelines for applying generative AI within intellectual property departments. I hope you will find the analysis informative.
Please note that the research and analysis generated by AI are based solely on publicly available information. They may not fully reflect actual circumstances and may contain inaccuracies, so they should be interpreted with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

AIエージェントが行う人間の指示に反する行動

21/7/2026

0 Comments

 
AIエージェントが会社の中で働くようになったとき、問題になるのは「仕事ができるか」だけではなく、人間の指示に反する行動を行うことです。命令に反対して実験を秘密裏に妨害する。ユーザーの不正に協力して記録を書き換える。AIの回答を採点する立場で、事実と逆の評価を出す。さらには、人間を実行役にして機密情報を外部へ伝えようとする――。
Anthropicが公開した研究では、Claude、GPT、Gemini、DeepSeek、Grokなど14の最先端AIを企業内エージェントとして動かし、意図的に問題を誘発したシミュレーションで、人間の指示に反する行動を行うことが起きる条件を調べた結果が報告されています。
生成AIにこの論文の内容を深掘りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
Agentic Misalignment in Summer 2026
Case studies of frontier models sabotaging code, assisting fraud, mislabeling, and coaching whistleblowers.
https://alignment.anthropic.com/2026/agentic-misalignment-summer-2026/
 
倫理観のある優等生だから嘘をつく=Anthropicが公開したAI監視の落とし穴
https://exawizards.com/column/ai-trend/news-07-16-2026/
 
【解説AI】優秀なAIほど、『自らの信念/倫理観』に則って嘘をつく可能性がある。Anthropic最新研究論文より
https://www.youtube.com/watch?v=2YJ6patq4-A&t=1492s
 
 
AI Agents Acting Contrary to Human Instructions
As AI agents begin working within organizations, the key concern is no longer simply whether they can perform their assigned tasks. Equally important is the possibility that they may act in ways that conflict with human instructions.
Examples include secretly sabotaging an experiment in opposition to management's directives, altering records to assist a user's misconduct, issuing evaluations that deliberately contradict the facts while serving as an evaluator of AI responses, or even attempting to use humans as unwitting intermediaries to transmit confidential information outside the organization.
A study published by Anthropic investigated the conditions under which such behaviors may emerge. The researchers evaluated 14 frontier AI models—including Claude, GPT, Gemini, DeepSeek, and Grok—operating as enterprise AI agents. By intentionally creating simulated scenarios designed to provoke problematic behavior, they examined the circumstances under which these AI systems could act contrary to human instructions.
I asked generative AI to conduct an in-depth analysis of this research paper, and I hope you will find the results informative. Please note that the investigation and analysis generated by AI are based solely on publicly available information. They may not necessarily reflect actual circumstances and may contain inaccuracies, so they should be interpreted with appropriate caution.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments

令和7年(行ケ)第10081号 手続違背

20/7/2026

0 Comments

 
令和7年(行ケ)第10081号は、知的財産高等裁判所で令和8年(2026年)7月8日に判決が言い渡された特許に関する審決取消請求事件で、発明の名称を「二酸化炭素を電気化学還元する方法」とする発明について、拒絶査定不服審判請求を不成立とした審決の取消訴訟において、審判手続で特許法159条2項、50条に基づき拒絶理由を通知し意見書を提出する機会が与えられなかったことが違法であるとして、審決が取り消された事例です。
原告は、取消事由として、審判手続において、拒絶理由に対して意見書及び補正書を提出する機会が与えられなかったとして、手続違背を主張した結果、本判決は、要旨以下の理由により、原告の主張する取消事由には理由があるとして、本件審決を取り消しました。
『本件拒絶査定と本件審決は、本願発明の「吸収剤」として異なる物質を認定したものであるところ、炭酸ナトリウム溶液又はアルカノールアミン溶液と「ギ酸とギ酸ナトリウムを含む混合溶液」とは、引用文献1において全く異なるプロセス及び目的で使用されており、同一のプロセスにおいて用いられる代替品や同等物といった関係にあるものではない。そうすると、「ギ酸とギ酸ナトリウムを含む混合溶液」を吸収剤と認定して本願を拒絶することは原告に対する不意打ちとなるから、原告にその旨を通知して意見書の提出及び補正の機会を与えるのが相当であって、「査定の理由と異なる拒絶の理由を発見した場合」(特許法159条2項)に該当するというべきである。そして、補正の機会を与えないという上記手続違背は、その性質上、本件審決結論に影響を及ぼすものと解される。したがって、本件審判手続には、特許法159条2項、50条に違反する違法があり、この点は本件審決を取り消すべき事由に当たると認められる。』
本判決について、生成AIに評釈させましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
令和8年7月8日判決言渡
令和7年(行ケ)第10081号 審決取消請求事件 判 決
https://www.courts.go.jp/assets/hanrei/hanrei-pdf-96784.pdf
 
 
IP High Court Case No. Reiwa 7 (Gyo-Ke) 10081: Procedural Violation
Case No. Reiwa 7 (Gyo-Ke) 10081 is an action seeking the revocation of a JPO trial decision concerning a patent application entitled "Method for Electrochemically Reducing Carbon Dioxide." The Intellectual Property High Court rendered its judgment on July 8, 2026. The case involved an appeal against a JPO trial decision that upheld the examiner's rejection of the patent application. The Court held that the trial proceedings were unlawful because the applicant had not been notified of a new ground for rejection or given an opportunity to submit arguments in an opinion or amend the claims, as required under Articles 159(2) and 50 of the Japanese Patent Act. Accordingly, the Court revoked the trial decision.
The plaintiff argued that the JPO trial proceedings were procedurally defective because it had been deprived of the opportunity to submit both an opinion and claim amendments in response to the newly adopted ground for rejection. The Court accepted this argument and revoked the trial decision for the following reasons:
"The examiner's final rejection and the JPO trial decision identified different substances as the 'absorbent' in the claimed invention. Sodium carbonate solution or alkanolamine solution and 'a mixed solution containing formic acid and sodium formate' are used in entirely different processes and for different purposes in Cited Reference 1. They are not substitutes or equivalents used within the same process. Therefore, treating 'a mixed solution containing formic acid and sodium formate' as the absorbent and rejecting the application on that basis constituted an unexpected change that unfairly surprised the applicant. The applicant should therefore have been notified of this new basis for rejection and given an opportunity to submit an opinion and amend the application. Accordingly, this situation falls within the meaning of 'where a ground for rejection different from the reasons for the examiner's decision is discovered' under Article 159(2) of the Patent Act. Furthermore, by its very nature, the procedural violation of denying the applicant an opportunity to amend could have affected the outcome of the trial decision. Consequently, the trial proceedings violated Articles 159(2) and 50 of the Patent Act, and this procedural defect constitutes grounds for revoking the trial decision."
I asked a generative AI system to prepare a commentary analyzing this judgment. Please note, however, that the AI's research and analysis are based solely on publicly available information. They may not necessarily reflect the full factual circumstances of the case and may contain inaccuracies. Please read the commentary with these limitations in mind.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments
<<Previous
Forward>>

    著者

    萬秀憲

    アーカイブ

    April 2026
    March 2026
    February 2026
    January 2026
    December 2025
    November 2025
    October 2025
    September 2025
    August 2025
    July 2025
    June 2025
    May 2025
    April 2025
    March 2025
    February 2025
    January 2025
    December 2024
    November 2024
    October 2024
    September 2024
    August 2024
    July 2024
    June 2024
    May 2024
    April 2024
    March 2024
    February 2024
    January 2024
    December 2023
    November 2023
    October 2023
    September 2023
    August 2023
    July 2023
    June 2023
    May 2023
    April 2023
    March 2023
    February 2023
    January 2023
    December 2022
    November 2022
    October 2022
    September 2022
    August 2022
    July 2022
    June 2022
    May 2022
    April 2022
    March 2022
    February 2022
    January 2022
    December 2021
    November 2021
    October 2021
    September 2021
    August 2021
    July 2021
    June 2021
    May 2021
    April 2021
    March 2021
    February 2021
    January 2021
    December 2020
    November 2020
    October 2020
    September 2020
    August 2020
    July 2020
    June 2020

    カテゴリー

    All

    RSS Feed

Copyright © よろず知財戦略コンサルティング All Rights Reserved.
サイトはWeeblyにより提供され、お名前.comにより管理されています
  • Home
  • Services
  • About
  • Contact
  • Blog
  • 知財活動のROICへの貢献
  • 生成AIを活用した知財戦略の策定方法
  • 生成AIとの「壁打ち」で、新たな発明を創出する方法