• 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

GPT-6 Astraより高性能な「社内モデル」

10/9/2026

0 Comments

 
2026年9月8日(現地時間)、OpenAIがミレニアム懸賞問題の一つであるナビエ・ストークス方程式の存在と滑らかさの問題に対する解法を公開しました。数学界では、この成果が本来の懸賞が問う「外力なし」のケースではない点や、第三者による査読が未完了である点等から、評価には慎重です。証明を生成したのは“GPT-6 Astraより大幅に高性能な”社内モデルということで、公開されているモデルより高性能な「社内モデル」が次の高性能な公開モデルを開発するパターンが生まれているようです。
生成AIに、この件について深堀調査を行わせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
On the Navier–Stokes Millennium Prize Problem
https://openai.com/index/navier-stokes-solution/
 
 
An “Internal Model” That Outperforms GPT-6 Astra
On September 8, 2026 (local time), OpenAI published a solution to the Navier–Stokes existence and smoothness problem, one of the Millennium Prize Problems. The mathematical community has been cautious in its assessment, however, noting, among other things, that the result does not address the “unforced” case specified in the original prize problem and that independent peer review has not yet been completed. The proof was reportedly generated by an internal model described as “significantly more capable than GPT-6 Astra.” A pattern therefore appears to be emerging in which internal models that outperform publicly available models are used to develop the next generation of more capable models for public release.
I asked generative AI to conduct an in-depth investigation into this topic, and invite you to review the findings. Please bear in mind that the AI-generated research and analysis are based solely on publicly available information, may not necessarily reflect the actual state of affairs, and could contain inaccuracies.

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

世界と日本のAI格差

10/9/2026

0 Comments

 
YouTubeで無料で公開されている動画『TBS CROSS DIG with Bloomberg「1on1 Tech」【世界と日本のAI格差】』は、収録日2026年9月1日に、Preferred Networks代表の岡野原大輔氏へインタビューしたもので、AI自身が次世代モデルを開発する自己改善ループの現状や、日米中の開発格差が語られています。日本が世界に追いつくための戦略として、新会社「ノエトラ」を拠点に国内企業が結集し、物理世界を理解する「フィジカルAI」の母体モデル構築を目指す方針などが語られています。
生成AIに、この動画を基に、世界と日本のAI格差について、追加の調査を行わせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
TBS CROSS DIG with Bloomberg「1on1 Tech」。
【世界と日本のAI格差】
https://www.youtube.com/watch?v=j6hdakNkiXk
 
 
September 10 — The AI Gap Between Japan and the Rest of the World
The video “The AI Gap Between Japan and the Rest of the World,” part of TBS CROSS DIG with Bloomberg’s 1on1 Tech series and available for free on YouTube, features an interview with Daisuke Okanohara, head of Preferred Networks, recorded on September 1, 2026. The discussion explores the current state of self-improvement loops in which AI itself develops next-generation models, as well as disparities in AI development among Japan, the United States, and China. As a strategy for helping Japan catch up with global leaders, it also covers plans to bring Japanese companies together around the new company Noetra to build a foundation model for “physical AI” capable of understanding the physical world.
Using this video as a starting point, I asked generative AI to conduct additional research into the AI gap between Japan and the rest of the world. Please refer to the findings. Please note, however, that the research and analysis produced by generative AI are based solely on publicly available information, may not necessarily reflect the actual state of affairs, and may contain inaccuracies. Please keep these limitations in mind when reviewing the findings.

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

生成AIの特許実務への利用可能性に関する検討

10/9/2026

0 Comments

 
生成AIの性能が急速に向上しています。専門家の論説を批評させることで、生成AIのレベルを把握すべく、月刊パテント 2026年7月号に掲載されている日本弁理士会令和6年度特許委員会生成AIワーキンググループが執筆した報告「生成AIの特許実務への利用可能性に関する検討」について、生成AIに批評させましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
生成AIの特許実務への利用可能性に関する検討
https://jpaa-patent.info/patent/viewPdf/4854
 
September 10 — Examining the Applicability of Generative AI to Patent Practice
Generative AI capabilities are improving rapidly. To gauge its performance through critiques of expert commentary, I asked generative AI to critically review the report “Examining the Applicability of Generative AI to Patent Practice,” written by the Generative AI Working Group of the Japan Patent Attorneys Association’s FY2024 Patent Committee and published in the July 2026 issue of the monthly journal Patent. Please take a look at the resulting critique.
Please note that the AI-generated research and analysis are based solely on publicly available information, may not accurately reflect actual circumstances, and may contain errors. Please read the critique 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

AIを発明者とする特許登録の可否について

9/9/2026

0 Comments

 
生成AIの性能が急速に向上しています。専門家の論説を批評させることで、そのレベルを把握すべく、月刊パテント 2026年7月号に掲載されている、日本弁理士会令和6年度特許委員会第1部会第2グループによる活動報告「AIを発明者とする特許登録の可否について」について、生成AIに批評させましたので、ご参照ください。
なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
AIを発明者とする特許登録の可否について
https://jpaa-patent.info/patent/viewPdf/4849
 
Can Patents Be Granted with AI Named as the Inventor?
The capabilities of generative AI are improving rapidly. To gauge the level it has reached, I asked it to critique a piece of expert commentary: the activity report “Can Patents Be Granted with AI Named as the Inventor?” The report was prepared by Group 2 of the First Subcommittee of the Japan Patent Attorneys Association’s FY2024 Patent Committee and published in the July 2026 issue of the monthly journal Patent. Please refer to the resulting critique.
Please note that the research and analysis produced by generative AI are based solely on publicly available information, do not necessarily reflect the actual circumstances, and may contain inaccuracies. Please keep these limitations in mind when reviewing the critique.
 

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

関西 知財・無形資産ガバナンス・フォーラム

8/9/2026

0 Comments

 
「関西 知財・無形資産ガバナンス・フォーラム~関西発 知財・無形資産ドリブン経営への変革~」が 知財・無形資産ガバナンス協会 と 大阪工業大学の共催で、2026年9月4日(金)に、大阪工業大学 梅田キャンパスで開催されました。
「関西 知財・無形資産ガバナンス・フォーラム」は、知的財産を単なる法的権利や費用ではなく、企業の持続的な価値を創造する成長投資と捉え直すための指針を示しました。
内閣府の最新政策やコーポレートガバナンス・コードの改訂を背景に、知財戦略を経営の核心に据える重要性が産官学の視点から説かれ、大手メーカーの組織運営や中小企業のヒット商品開発におけるMPDP理論など、知財をビジネスに結びつける具体的な実践事例が豊富に紹介されました。また、生成AI時代の到来を見据え、専門知識だけでなく経営の視点や対話力を兼ね備えた次世代の人財育成が不可欠であると強調、本フォーラムは、知財を基軸とした日本の経済再生に向け、現場の行動変革と価値創造ストーリーの構築を促す強力なメッセージを発信しました。
 
講演 ① 「知的財産推進計画2026 ~成長戦略を支える知財戦略の推進~」 内閣府知的財産戦略推進事務局 参事官 清水 祐樹 様
講演 ② 「MPDP理論による中小企業活性化戦略」 株式会社エンジニア 代表取締役社長 高崎 充弘 様
講演 ③ 「経営に活かす知財の在り方を考える ~クリエイティブな知財部運営に挑むダイキン工業の取組み~」 ダイキン工業株式会社 知的財産部長 安部 剛夫 様
講演 ④ 「知的資産と知財教育」 大阪工業大学 知的財産学部 杉浦 淳 様
パネル ディスカッション 「知財を経営に活かす人財」のあるべき姿 モデレータ:KDDI株式会社 川名 弘志 氏(知財・無形資産ガバナンス協会 副理事長) パネラー :清水 祐樹 様 / 高崎 充弘 様 / 安部 剛夫 様 / 杉浦 淳 様
 
以下は、講演の私的メモです。
① 知財・無形資産を「費用」ではなく「資産形成」と捉える――清水祐樹氏の講演
9月4日の「関西 知財・無形資産ガバナンス・フォーラム」では、内閣府知的財産戦略推進事務局の清水祐樹氏が、知的財産推進計画2026と、成長戦略を支える知財政策について講演しました。
講演の中で特に重要だと感じたのは、知財・無形資産への支出を、単なる費用ではなく、将来の価値を生み出す資産形成として考えるという視点です。清水氏は、ガイドラインの考え方として、知財を保有するだけでなく、価格決定力や競争のルールを変える力につなげることを説明しました。その実践には、現状の把握、重要課題の特定、戦略上の位置付けの明確化、価値創造ストーリーの構築が必要であり、投資家や金融機関との対話も重要になるとしています。
2026年のコーポレートガバナンス・コード改訂については、知財への投資が成長投資として取締役会の役割に位置付けられたと説明されました。また、知財の創出・取得・強化・保護・収益化に戦略的に取り組む考え方を踏まえ、ガイドラインの改訂を進めていることも紹介されました。講演時点では年度末までの改訂を目指しており、それに先立つ実践の指針として、8月3日にガイダンスを公表したとの説明でした。
生成AIについては、利用を進めることと知財を保護することを、どう両立させるかが取り上げられました。技術的に利用できても、法的・倫理的な不安があれば活用は進みません。そこで、AI事業者による情報開示や問い合わせへの対応を促すプリンシプル・コードの考え方が説明されました。原則に対応するか、対応できない理由を説明する仕組みを通じて、透明性を高めようとするものです。
研究開発との関係も重要な論点でした。研究成果を論文として発表してから特許を考えるのではなく、初期段階から出願時期や知財の扱いを検討する。同様に、国際標準が事業展開の障害になってから対応するのではなく、研究開発・知財・標準・事業戦略を一体で考える必要性が示されました。何をオープンにして普及を進め、何をクローズにして競争優位を確保するかという判断です。
この講演から受け取ったのは、知財政策を「制度への対応」として読むだけでは不十分だということです。自社の成長に向けて、何に投資し、何を蓄積し、どう活用するか。その議論の中に知財を位置付けてこそ、政策と企業実務がつながります。知財戦略を「出願の計画」から、「事業の成長を支える投資と活用の計画」へ広げる必要性を感じる講演でした。
 
② ネジザウルスの成功に学ぶMPDP――特許だけでは商品は売れない
9月4日の「関西 知財・無形資産ガバナンス・フォーラム」で、株式会社エンジニアの高崎充弘氏が、「MPDP理論による中小企業活性化戦略」について講演しました。
高崎氏の話は、ネジザウルスGTの成功を、単なるヒット商品の物語で終わらせないものでした。それ以前の商品と何が違ったのかを振り返り、マーケティング、パテント、デザイン、プロモーションという四つの要素を抽出したのがMPDPです。かつては、特許を取れば売れる、デザインを良くすれば売れると考えていたものの、一つの要素だけでは十分ではなかった。四つが重なり合うところを狙うことが重要だと気付いたという説明に、長年の商品開発の経験が表れていました。
印象的だったのは、顧客アンケートに寄せられた少数の要望を商品化につなげたエピソードです。グリップなどの改良要望に交じって、「トラスねじも外したい」という声がありました。社内では採用するか意見が分かれましたが、実際に対応すると、この機能が顧客に強く響いたといいます。多くの人が口にする不満だけでなく、まだ十分に言葉になっていない潜在ニーズを見極めることの重要性が示されました。ただし、高崎氏は、少数意見のすべてが大きな需要につながるわけではなく、その見極めが難しいことにも触れています。
デザインやプロモーションにも具体的な工夫がありました。展示会では、リアルな恐竜が一部の来場者に敬遠されたことから、キャラクターを使った展開に切り替えたと紹介されました。また、斬新な商品ほど、その新しさに見合う説明や訴求が必要だと指摘しています。新しいものを作る力と、その価値を顧客に伝える力は、切り離せないということです。
中小企業の知財活用については、「二階からビール」という比喩が紹介されました。専門家から有益な助言を注いでもらっても、受け手である経営者に基礎知識がなければ、コップで受け止めきれない。高崎氏は、自ら知財の検定を学ぶことで制度の全体像が分かり、専門家との意思疎通が改善した経験を語りました。必要なのは、経営者が弁理士と同じ専門家になることではなく、相談し、理解し、判断するための共通言語を持つことです。社員の学習にも、その考え方を広げていると説明されました。
終盤には、医療用への展開や、ねじの写真をAIで解析して対応を案内する構想も紹介されました。顧客の困りごとを解決する活動を広げ、製品ブランドと企業ブランドを育てていくという方向性です。
この講演は、知財支援を「特許取得の支援」だけで終わらせてはいけないことを、改めて考えさせるものでした。売れる価値を見つけ、守り、形にし、伝える。その一連の活動の中で知財を使いこなすことが、中小企業の経営を強くするのだと感じます。
 
③ 知財部の仕事を「開発行為」として捉える――ダイキン工業の実践
9月4日の「関西 知財・無形資産ガバナンス・フォーラム」で、ダイキン工業の安部剛夫氏が、経営に活かす知財と、クリエイティブな知財部運営について講演しました。
講演を象徴するのが、知財部の仕事そのものを「価値創造プロセスにおける一つの開発行為」として捉えるという考え方です。出願・権利化を重要な基盤としながら、蓄積された知をどう価値に変えるか、そのためにどのような課題を解決するかを考える。知財部が扱う仕事の幅を広げるだけでなく、仕事に向き合う姿勢そのものを変えようとする説明でした。
IPランドスケープについても、分析資料を作ること自体が目的ではありません。安部氏は、現在の状況や将来の可能性を把握する「解像度」を高める役割として説明しました。戦略立案、特許ポートフォリオの形成、外部との連携、契約交渉など、さまざまな活動を適切な方向に進めるための機能という位置付けです。
その前提として重視しているのが、現場とのコミュニケーションです。現場の課題を聞かずに分析を進めると、方向がずれてしまう。何を実現したいのか、本当の目的は何かを掘り下げたうえで、知財部からも提案していくと説明されました。「IPインテリジェンス」という言葉から高度な分析ツールを思い浮かべがちですが、その出発点は人との対話にあるという指摘が印象的でした。
事業と連動した知財活動の例としては、R32冷媒に関する特許開放が紹介されました。先行して技術開発と特許取得を進めても、その特許が普及の障害になれば市場は広がりません。そこで、標準化の取り組みとともに関連特許を開放し、普及を進める戦略を取ったという説明です。また、大学やベンチャーとの協創では、相手の事情を聞き、双方に利益がある契約を設計することを重視していると述べました。
生成AIの活用も、こうした役割の拡大と結び付いていました。安部氏は、壁打ち、大量処理、検討内容の磨き上げなどへの有用性を説明する一方、構想の出発点や最終的な仕上げでは、人との役割分担が必要とする見方を示しました。特許分類やFTO業務の一部、特許マップなどへの活用も紹介されています。重要なのは、効率化によって、出願・権利化に投入していたリソースを、IPLや外部との協創、事業に連動した活動へ振り向けるという考え方です。
組織面では、本社と研究開発拠点の双方に関わり、権利化と技術開発の両方の現場感を持てる体制づくりが紹介されました。現場に入り、課題や目的を正確に把握し、行動につなげるための取り組みです。安部氏は、まだ試行錯誤の途中であることも率直に語っています。
私がこの講演から特に考えさせられたのは、AI活用の成果を時間短縮だけで測らないことです。効率化によって生まれた余力で、知財部はどの事業課題を解決するのか。 そこまで考えて初めて、AI活用と知財部門の変革が一つにつながるのではないでしょうか。
 
④ 知財教育を、事業と社会で活きる「実践」へ――杉浦淳氏の講演
9月4日の「関西 知財・無形資産ガバナンス・フォーラム」で、大阪工業大学の杉浦淳氏が、「知的資産と知財教育」について講演しました。
講演前半では、知財立国政策と人材育成の歩みを振り返り、知識や情報が経済・企業の競争力を支えるという議論が紹介されました。さらに、創作を収入に結び付け、創作者が自らの創造によって生活できるようにするという知財の役割が取り上げられました。一方で、知財制度の効果や社会的影響を疑問視する議論にも触れており、知財の意義を経済と社会の双方から考えさせる内容でした。
そのうえで、教育の具体的な取り組みとして繰り返し語られたのが、「実践」です。知財法をしっかり学ぶことを基礎に、経営、事業戦略、競争戦略、知財の価値評価、知財を利用した事業化演習へと学習を広げています。ケーススタディでも、制度上の結論だけでなく、事業の立ち位置、競争相手、顧客との関係、差別化の方法などを考えると説明されました。
学んだ内容を実際に使う機会として、学生のコンテストへの参加も紹介されました。知識を覚えるだけでなく、具体的な提案にまとめ、外部の評価を受ける。そこまで経験させることを大切にしているという説明です。
生成AIを使った分野横断型の学習も興味深い事例でした。知財・情報・デザインを学ぶ学生が、フィリピンの学生も交えてチームを組み、3日間の試作を中心とした学習に取り組んだと紹介されました。現地の文化や市場を考えながらゲームの案を作り、発表するというものです。創造、保護、ビジネスを別々の科目として学ぶだけでなく、一つの活動として経験する場になっていました。
AI時代の知財については、従来業務の多くが置き換わっていくとの見通しを踏まえ、さらに研究開発や事業経営と一体となった活動へ進む必要性が語られました。業務を効率化するだけでなく、知財を使って何を実現するのかを考えることが、教育にも求められていると受け止めました。
講演の視野は、企業経営だけにとどまりません。地域の伝統野菜など、みんなに関わる資産をどう維持・活用するかという問題や、タイ北部でGIを活用して地域の仕事を支える事例も紹介されました。また、高校段階から知財教育を広げ、担い手の裾野を厚くする必要性も示されました。
この講演から、知財教育は「制度を理解する人」を育てるだけでなく、創造を事業や地域の暮らしにつなげる人を育てる営みでもあると感じました。知財の専門性を土台に、異なる分野の人と協力し、実際に価値を生み出す。その力をどう育てるかが、これからの知財教育の重要な課題なのだと思います。
 
⑤ 経営に生きる知財人材とは――専門性と当事者意識をどうつなぐか
「関西 知財・無形資産ガバナンス・フォーラム」のパネルディスカッションでは、川名弘志氏をモデレーターに、4講演の登壇者が「経営に生きる知財人材とは何か」を議論しました。法律・技術の専門家にとどまらず、経営陣や事業部門と同じ目線で、企業価値の向上に関わるには何が必要なのかという問いです。
冒頭で川名氏は、協会の人材育成プログラムを紹介しました。役割の変化を認識するところから始め、投資家・経営者の視点を学び、知財業務での実践、組織・人材の問題へとつなげる構成です。最後には受講者がアクション宣言を行い、半年後に進捗を共有します。学んで終わりではなく、行動を変えるところまで支援するという姿勢が示されました。
清水氏が指摘したのは、経営層に伝えるための経営知識と、「なぜその知財が重要なのか」を問い直す姿勢です。知財に携わる側は、自分たちが作った知財を重要だと考えがちです。しかし、その前提を当然とせず、全社を俯瞰し、経営の文脈で説明する必要があると述べました。
高崎氏は、中小企業にはそもそも知財の専門家が社内にいない場合があるという、出発点の違いを示しました。そのため、開発や営業の担当者が、日々の仕事の中で権利化の可能性や他社権利への注意点に気付けることが重要になります。さらに、マーケティング、パテント、デザイン、プロモーションをつなぎ、商品化全体を動かす「MPDPのプロデューサー」を育てたいと語りました。
安部氏は、高い専門性、早い段階で情報を得る力、コミュニケーション力に加え、最後まで具体化する実行力を挙げました。コメントをするだけではなく、事業に伴走し、作り切るところまで関わる必要があるという指摘です。
さらに印象的だったのが、「主となる座標はビジネスであり、知財はそれを支え、活かすための座標である」という説明です。専門性を鍛えているうちに、知財そのものが目的になってしまうことがある。だからこそ、事業への当事者意識を失わず、自分たちはどの座標で考えているのかを、対話によって問い直す必要があると述べました。
ただし、全員を同じ人材像に変える必要はないという点も重要です。専門性に非常に強い人と、事業の視点に強い人を組み合わせ、それぞれの強みを活かせばよい。知財側が経営を学ぶことと、経営側が知財を学ぶことの両方が必要だという整理でした。
杉浦氏は、法律実務を基礎に経営と知財を結び付け、さらに「経営から知財を見る」教育へ進む考え方を示しました。産官学連携についても、学校や学生の状況に合わせた知財教育を、民間の実務経験も活かしながら組み立てる必要性を述べています。
この討論から見えてきたのは、「経営に活きる知財人材」を、一人で何でもできる万能な専門家として考える必要はないということです。専門性を高めること、事業の当事者として考えること、異なる強みを持つ人と協力して実行すること。 これらを個人と組織の両方で実現していくことが、知財を経営の力に変える道筋なのだと感じました。
 
 
September 8 — Kansai IP & Intangible Assets Governance Forum
The “Kansai IP & Intangible Assets Governance Forum—Driving the Transformation to IP- and Intangible Asset-Driven Management from Kansai” was held on Friday, September 4, 2026, at the Umeda Campus of Osaka Institute of Technology. The event was jointly organized by the Intellectual Property and Intangible Assets Governance Association and Osaka Institute of Technology.
The forum offered guidance on reframing intellectual property not merely as legal rights or a cost, but as an investment in growth that creates lasting corporate value.
Against the backdrop of the Cabinet Office’s latest policies and revisions to Japan’s Corporate Governance Code, speakers from industry, government, and academia explained the importance of placing IP strategy at the heart of business management. Numerous practical examples illustrated how to connect IP with business, ranging from organizational management at major manufacturers to the application of the MPDP framework in developing successful products at small and medium-sized enterprises (SMEs). Looking ahead to the generative AI era, speakers also emphasized the importance of developing the next generation of professionals who combine specialist expertise with a management perspective and strong communication skills. The forum delivered a powerful message encouraging changes in day-to-day practice and the development of value creation narratives to support Japan’s economic revitalization through IP.
Lecture 1: “Intellectual Property Strategic Program 2026—Advancing IP Strategies to Support Growth Strategies”
Yuki Shimizu
Counsellor, Secretariat of Intellectual Property Strategy Headquarters, Cabinet Office
Lecture 2: “Strategies for Revitalizing SMEs through the MPDP Framework”
Mitsuhiro Takasaki
President and Representative Director, Engineer Inc.
Lecture 3: “Rethinking the Role of IP in Business Management—Daikin Industries’ Efforts to Build a Creative IP Department”
Takeo Abe
General Manager, Intellectual Property Department, Daikin Industries, Ltd.
Lecture 4: “Intellectual Assets and IP Education”
Jun Sugiura
Faculty of Intellectual Property, Osaka Institute of Technology
Panel Discussion: “The Ideal Profile of Professionals Who Put IP to Work in Business Management”
Moderator: Hiroshi Kawana, KDDI Corporation; Vice Chair, Intellectual Property and Intangible Assets Governance Association
Panelists: Yuki Shimizu, Mitsuhiro Takasaki, Takeo Abe, and Jun Sugiura
The following are my personal notes from the forum.
1. Viewing Spending on IP and Intangible Assets as “Building Assets,” Not Simply “Incurring Costs”—Yuki Shimizu’s Lecture
At the Kansai IP & Intangible Assets Governance Forum on September 4, Yuki Shimizu of the Cabinet Office’s Secretariat of Intellectual Property Strategy Headquarters spoke about the Intellectual Property Strategic Program 2026 and IP policies that support growth strategies.
What I found particularly important was the perspective that spending on IP and intangible assets should be viewed not simply as an expense, but as the building of assets that generate future value. Explaining the thinking behind the guidelines, Shimizu emphasized going beyond merely holding IP to using it to gain pricing power and the ability to reshape the rules of competition. Putting this into practice requires assessing the current situation, identifying key issues, clarifying IP’s strategic position, and constructing a value creation narrative. He also highlighted the importance of dialogue with investors and financial institutions.
Regarding the 2026 revision of the Corporate Governance Code, he explained that IP investment had been positioned as growth investment within the responsibilities of the board of directors. He also introduced ongoing revisions to the guidelines reflecting a strategic approach to creating, acquiring, strengthening, protecting, and monetizing IP. At the time of the lecture, the aim was to complete the revisions by the end of the fiscal year. He explained that guidance had been published on August 3 to provide practical direction ahead of those revisions.
The discussion of generative AI addressed how to reconcile promoting its use with protecting IP. Even when a technology is available for use, adoption will not advance if legal and ethical concerns remain. In this context, Shimizu explained the concept of a Principles Code that encourages AI providers to disclose information and respond to inquiries. It seeks to improve transparency through a mechanism under which providers either comply with the principles or explain why they cannot do so.
The relationship with research and development was another important topic. Rather than considering patents only after publishing research findings in academic papers, organizations need to consider filing timelines and the treatment of IP from the earliest stages. Likewise, rather than responding only after international standards become obstacles to business expansion, they need to consider R&D, IP, standardization, and business strategy together. This involves deciding what to make open to encourage adoption and what to keep closed to secure a competitive advantage.
My takeaway from this lecture was that it is not enough to read IP policy simply as a matter of responding to the institutional framework. What should a company invest in, what should it accumulate, and how should it use those assets to support its growth? Policy becomes connected to corporate practice only when IP is situated within that discussion. The lecture underscored the need to broaden IP strategy from “a plan for filing applications” to “a plan for investment and utilization that supports business growth.”
2. Learning MPDP from the Success of Neji-Saurus—Patents Alone Do Not Sell Products
At the Kansai IP & Intangible Assets Governance Forum on September 4, Mitsuhiro Takasaki of Engineer Inc. delivered a lecture on “Strategies for Revitalizing SMEs through the MPDP Framework.”
Takasaki’s account went beyond simply telling the story of Neji-Saurus GT as a successful product. By looking back at what distinguished it from earlier products, he identified four elements: Marketing, Patent, Design, and Promotion. These form the MPDP framework. He explained that he had once believed that obtaining a patent would make a product sell, or that improving its design would make it sell. Yet no single element was sufficient. His realization that it is important to target the intersection of all four elements reflected many years of product development experience.
One particularly memorable episode concerned turning requests from a small number of customers into a product feature. Among customer survey responses calling for improvements to the grips and other features was a request to “remove truss-head screws as well.” Opinions within the company were divided over whether to address this request, but once the capability was incorporated, it resonated strongly with customers. The example illustrated the importance of identifying not only widely expressed frustrations, but also latent needs that have not yet been fully articulated. Takasaki also noted, however, that not every request voiced by a small number of customers translates into substantial demand, and that distinguishing those that do is difficult.
He also shared specific approaches to design and promotion. At trade shows, realistic dinosaur displays had put off some visitors, prompting a shift to a mascot-based approach. He also pointed out that the more novel a product is, the more it needs explanations and promotional messages that do justice to its novelty. The ability to create something new and the ability to communicate its value to customers are inseparable.
In discussing IP utilization by SMEs, Takasaki introduced the metaphor of “pouring beer from the second floor.” Even when experts pour out useful advice, business owners without basic knowledge cannot properly catch it in their glasses. He recounted how studying for an IP certification exam himself helped him understand the overall system and communicate more effectively with experts. Business owners do not need to become specialists on a par with patent attorneys. They need a shared language that enables them to consult, understand, and make decisions. He explained that he is extending this approach to employee learning as well.
Toward the end of the lecture, he also discussed expansion into medical applications and a concept for using AI to analyze photographs of screws and guide users toward appropriate solutions. The direction is to broaden efforts to solve customers’ problems while strengthening both the product brand and the corporate brand.
This lecture reminded me that IP support should not end with helping companies obtain patents. The task is to identify value that customers will pay for, protect it, give it tangible form, and communicate it. I felt that putting IP to work throughout this entire sequence of activities is what makes SMEs stronger businesses.
3. Viewing the Work of the IP Department as a “Development Activity”—Daikin Industries’ Approach
At the Kansai IP & Intangible Assets Governance Forum on September 4, Takeo Abe of Daikin Industries spoke about putting IP to work in business management and running a creative IP department.
The idea that best captured his lecture was that the work of the IP department itself should be viewed as “a development activity within the value creation process.” While recognizing patent filing and the securing of rights as an essential foundation, the department considers how to turn accumulated knowledge into value and what challenges must be solved to achieve that. His explanation was not simply about broadening the scope of the department’s work, but about changing its fundamental approach to that work.
Similarly, the purpose of IP landscape analysis is not simply to produce analytical reports. Abe described its role as improving the “resolution” with which the company understands its current situation and future possibilities. It is a function that helps guide a variety of activities in the right direction, including strategy formulation, patent portfolio development, external collaboration, and contract negotiations.
An essential foundation for this is communication with the people directly involved in the work. Analysis can head in the wrong direction if it proceeds without first listening to the challenges they face. Abe explained that the IP department seeks to understand what those colleagues want to accomplish and what their underlying objectives really are, and then makes proposals of its own. The term “IP intelligence” may call to mind sophisticated analytical tools, but I was struck by his point that its starting point is dialogue with people.
As an example of IP activities aligned with business strategy, Abe discussed opening up patents relating to R32 refrigerant. Even if a company leads the way in technology development and patent acquisition, the market will not expand if those patents become barriers to adoption. He explained that Daikin therefore adopted a strategy of opening up relevant patents alongside its standardization efforts to encourage wider adoption. In co-creation with universities and startups, he also emphasized listening to the other party’s circumstances and designing agreements that benefit both sides.
The use of generative AI was also connected to this expansion of the department’s role. Abe described its usefulness as a sounding board, for processing large volumes of information, and for refining ideas under consideration. At the same time, he expressed the view that an appropriate division of roles between people and AI is necessary at the initial concept stage and in the final refinement of the work. Examples of AI use included patent classification, parts of freedom-to-operate (FTO) work, and patent mapping. The key is to use efficiency gains to redirect resources previously devoted to patent filing and securing rights toward IP landscape analysis (IPL), co-creation with external partners, and activities aligned with the business.
On the organizational side, Abe described efforts to build a structure that enables personnel to engage with both headquarters and R&D sites and develop firsthand understanding of both rights acquisition and technology development. These efforts are intended to help the department work directly with the relevant teams, accurately understand their challenges and objectives, and translate that understanding into action. He also spoke candidly about the fact that this remains a process of trial and error.
What this lecture particularly prompted me to consider was the importance of not measuring the benefits of AI solely in terms of time saved. Which business challenges will the IP department solve with the capacity freed up by greater efficiency? It seems to me that AI adoption and the transformation of the IP function become truly connected only when we think that far.
4. Turning IP Education into Practice That Matters in Business and Society—Jun Sugiura’s Lecture
At the Kansai IP & Intangible Assets Governance Forum on September 4, Jun Sugiura of Osaka Institute of Technology delivered a lecture on “Intellectual Assets and IP Education.”
In the first half of the lecture, Sugiura reviewed the history of Japan’s policy of becoming an IP-based nation and its efforts to develop the necessary human resources. He introduced the argument that knowledge and information underpin economic and corporate competitiveness. He also discussed IP’s role in connecting creative work to income so that creators can make a living from their own creations. At the same time, he touched on arguments questioning the effectiveness and social impact of the IP system, prompting reflection on its significance from both economic and social perspectives.
Against this background, the word that repeatedly emerged in his discussion of specific educational initiatives was “practice.” Building on a solid grounding in IP law, students broaden their learning to encompass management, business strategy, competitive strategy, IP valuation, and practical exercises in commercialization using IP. He explained that case studies consider not only conclusions under the legal framework, but also the position of the business, its competitors, customer relationships, and ways to differentiate.
Sugiura also introduced students’ participation in competitions as an opportunity to apply what they have learned. Rather than merely memorizing knowledge, students develop concrete proposals and submit them to external evaluation. He explained the importance of giving students experience of that entire process.
An interesting example was interdisciplinary learning using generative AI. Students studying IP, information science, and design formed teams that also included students from the Philippines and took part in a three-day learning program centered on prototyping. They developed and presented game concepts while considering local culture and markets. Rather than learning about creation, protection, and business only as separate subjects, students had an opportunity to experience them as an integrated activity.
Regarding IP in the AI era, Sugiura discussed the prospect that many conventional tasks will be replaced by AI and the resulting need for IP activities to become more closely integrated with R&D and business management. I understood this to mean that education, too, must go beyond improving operational efficiency and encourage people to consider what they want to achieve through IP.
The scope of the lecture extended beyond corporate management. Sugiura discussed how to maintain and use assets of shared importance, such as traditional local vegetable varieties, and introduced an example from northern Thailand in which geographical indications (GIs) are used to support local livelihoods. He also emphasized the need to expand IP education beginning at the high school level and broaden the pool of people capable of putting IP to use.
This lecture made me feel that IP education is not only about developing people who understand the system. It is also about nurturing people who connect creativity with business and the lives of local communities. Building on specialist IP expertise, they need to collaborate with people in other fields and create real value. How to cultivate that capacity is, I believe, a key challenge for IP education in the years ahead.
5. What Makes IP Professionals Valuable to Business?—Connecting Expertise with a Sense of Ownership
In the panel discussion at the Kansai IP & Intangible Assets Governance Forum, Hiroshi Kawana moderated a discussion among the four lecturers about what makes an IP professional who contributes to business management. The central question was what it takes to move beyond being a legal or technical specialist and contribute to enhancing corporate value from the same perspective as senior management and business divisions.
At the outset, Kawana introduced the association’s professional development program. It begins by helping participants recognize changes in their roles, then moves on to learning the perspectives of investors and business leaders, applying that learning in IP work, and addressing organizational and people-related issues. At the end, participants make a commitment to action and share their progress six months later. The approach is designed not to end with learning, but to support changes in behavior.
Shimizu emphasized the management knowledge needed to communicate with senior executives and the willingness to reconsider “why this IP is important.” People involved in IP tend to regard the IP they have created as important. Rather than taking that assumption for granted, however, he said that they need to view the company as a whole and explain its significance in the context of management.
Takasaki highlighted a different starting point for SMEs: some do not have any in-house IP specialists at all. It is therefore important for people in development and sales to recognize opportunities to secure rights and potential issues involving other companies’ rights in their everyday work. He also expressed his desire to develop “MPDP producers” who can connect marketing, patents, design, and promotion and drive the entire commercialization process.
Abe identified deep expertise, the ability to obtain information at an early stage, communication skills, and the ability to execute and bring ideas to fruition. His point was that IP professionals need to do more than offer comments. They must work alongside the business and remain involved through to completion.
Particularly memorable was his explanation that “business is the primary frame of reference, while IP provides a frame of reference for supporting it and realizing its potential.” As professionals deepen their expertise, IP itself can sometimes become the objective. For that reason, he said, they need to retain a sense of ownership of the business and use dialogue to reconsider the frame of reference from which they are thinking.
An equally important point was that not everyone needs to be molded into the same type of professional. People with exceptional specialist expertise can be paired with those who have strong business perspectives, allowing each to contribute their strengths. The discussion emphasized that learning must go both ways: IP professionals need to learn about management, and business leaders need to learn about IP.
Sugiura described an educational approach that builds on legal practice, connects management and IP, and then goes further to teach people to “look at IP from a management perspective.” On collaboration among industry, government, and academia, he also emphasized the need to design IP education suited to the circumstances of schools and students while drawing on practical experience from the private sector.
What emerged from this discussion was that professionals who put IP to work in business do not have to be all-purpose experts capable of doing everything themselves. They need to deepen their expertise, think as people with a direct stake in the business, and work with others who bring different strengths to get things done. I felt that achieving this at both the individual and organizational levels is the path to turning IP into a source of business strength.

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

Artificial Analysis Intelligence Index v4.2公表

7/9/2026

0 Comments

 
OpenAI「GPT-6 Astra」の評価が、「Epoch AIの総合評価」や「ARC-AGI-3」では高かったのに対し、「Artificial Analysis Intelligence Index」の評価では「GPT-5.6 Sol」と同水準だったことには違和感がありました。
そういう声が大きかったのか、2026年9月4日、LLM総合評価指標「Artificial Analysis Intelligence Index」のv4.2が公表されました。
この改定について、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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Artificial Analysis「Announcing Artificial Analysis Intelligence Index v4.2」
https://artificialanalysis.ai/articles/artificial-analysis-intelligence-index-v4-2
 
 
Announcing Artificial Analysis Intelligence Index v4.2
I found it puzzling that OpenAI’s GPT-6 Astra received high ratings in Epoch AI’s overall evaluation and on ARC-AGI-3, yet scored on par with GPT-5.6 Sol on the Artificial Analysis Intelligence Index.
Perhaps in response to a growing chorus of such concerns, version 4.2 of the Artificial Analysis Intelligence Index, a composite index for evaluating large language models (LLMs), was released on September 4, 2026.
I asked generative AI to conduct an in-depth examination of this revision, and the findings are provided for your reference. Please bear in mind, however, that the AI-generated research and analysis are based solely on publicly available information, may not necessarily reflect the actual situation, and may contain 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.
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

米国勢4社が相次いで次世代基盤モデルを発表

7/9/2026

0 Comments

 
2026年9月1日から3日にかけてAnthropic、Google、Meta、OpenAIの4社が相次いで次世代基盤モデルを発表しました。Claude Fable 5.1、Gemini 3.8 Flash、Muse Spark 1.3,
 GPT-6 Astraで、Claude Fable 5.1は総合知能と科学研究支援で首位を維持する一方、GPT-6 Astraは数学やコンピュータ操作において圧倒的な優位性を示していると分析されています。また、軽量モデルであるGemini 3.8 Flashはマルチモーダル処理能力、Muse Spark 1.3は優れたコード修正精度に特徴があり、今後は用途に応じた使い分けが進むものと考えられます。
生成AIにこの4つについて深堀りさせましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Four U.S. Companies Announce Next-Generation Foundation Models in Rapid Succession
Between September 1 and 3, 2026, Anthropic, Google, Meta, and OpenAI announced their next-generation foundation models in rapid succession: Claude Fable 5.1, Gemini 3.8 Flash, Muse Spark 1.3, and GPT-6 Astra, respectively.
Analyses suggest that Claude Fable 5.1 continues to lead in overall intelligence and support for scientific research, while GPT-6 Astra demonstrates an overwhelming advantage in mathematics and computer use. Meanwhile, the lightweight Gemini 3.8 Flash stands out for its multimodal processing capabilities, and Muse Spark 1.3 for the accuracy of its code fixes. These differences suggest that users will increasingly choose models according to their specific use cases.
I asked generative AI to conduct an in-depth investigation of these four models and invite you to review the findings. Please note, however, that the AI-generated research and analysis are based solely on publicly available information, may not necessarily reflect the actual state of affairs, and could contain inaccuracies. Please keep these limitations in mind when reviewing the findings.

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

GPT-6 Astraは「サイエンスエージェント化」したか?

7/9/2026

0 Comments

 
2026年9月に相次いで発表された最先端AIモデル、GPT-6 AstraとClaude Fable 5.1(およびMythos 5.1)は、単なる知識の要約を超え、自律的にツールを操作して研究サイクルを回す「サイエンスエージェント」としての質的転換を果たしたと評価されています。
GPT-6 Astraは、数学的難問の解決やOSレベルの高度な操作性において圧倒的な実務遂行能力を示す一方、Claude Fableシリーズは学際的な知見の統合や実験データの解析、外部ラボと連携した分子設計などの理論探究に強みを持っています。
GPT-6 AstraとClaude Fable 5.1(およびMythos 5.1)の科学研究能力について、生成AIに比較・分析させましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Has GPT-6 Astra Become a “Science Agent”?
The cutting-edge AI models GPT-6 Astra and Claude Fable 5.1 (along with Mythos 5.1), announced in rapid succession in September 2026, are regarded as having undergone a qualitative transformation into “science agents”—moving beyond merely summarizing knowledge to autonomously operating tools and carrying out research cycles.
While GPT-6 Astra demonstrates exceptional task-execution capabilities in solving challenging mathematical problems and performing sophisticated operations at the operating-system level, the Claude Fable series shows strengths in theoretical exploration, including integrating interdisciplinary knowledge, analyzing experimental data, and designing molecules in collaboration with external laboratories.
I asked generative AI to compare and analyze the scientific research capabilities of GPT-6 Astra and Claude Fable 5.1 (along with Mythos 5.1), and am sharing the findings for your reference. Please note, however, that the AI-generated research and analysis are based solely on publicly available information, do not necessarily reflect the actual state of affairs, and may contain inaccuracies. Please review the findings 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.
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

GPT-6 Astraの登場で知財業務はどう変わるか

7/9/2026

0 Comments

 
OpenAIが2026年9月3日に発表したGPT-6 Astraは、「ARC-AGI-3」では、Standard 62.7%、 Provider Adapter 99.9%と、これまでの水準を大きく超えるなど、Computer Use、Browsing、Software Engineering、Science、Professional Workなどで高い性能を持ち、コード、ブラウザ、専門ソフトをまたぐ複数ステップの作業を得意としています。
この「GPT-6 Astra」の登場によって知財業務はどう変わるか、について生成AIに深掘り調査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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
A new generation of intelligence
https://openai.com/index/gpt-6-astra/
 
 
How Will the Arrival of GPT-6 Astra Change Intellectual Property Work?
GPT-6 Astra, announced by OpenAI on September 3, 2026, achieved scores of 62.7% (Standard) and 99.9% (Provider Adapter) on ARC-AGI-3, far surpassing previous performance levels. It demonstrates strong capabilities in areas such as computer use, web browsing, software engineering, science, and professional work, and excels at multi-step tasks spanning code, browsers, and specialized software.
I asked generative AI to conduct an in-depth investigation into how the arrival of GPT-6 Astra will change intellectual property work. Please take a look at the findings.
When reviewing these findings, please bear in mind that AI-generated research and analysis are based solely on publicly available information, do not necessarily reflect real-world conditions, and may contain 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.
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

GPT-6 Astraの初期評価

7/9/2026

0 Comments

 
2026年9月3日に発表されたGPT-6 Astraを有料ユーザーが使えるようになりましたので、使った人の評価・評判を、特に、GPT-5.6との違いに焦点をあてて、生成AIに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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Early Reviews of GPT-6 Astra
GPT-6 Astra, announced on September 3, 2026, is now available to paid users. I asked generative AI to research reviews and impressions from people who have used the model, focusing particularly on how it differs from GPT-5.6. Please take a look at the findings.
Please note that the AI-generated research and analysis are based solely on publicly available information and may not necessarily reflect the actual situation. They may also contain errors, so please bear these limitations in mind when reviewing the findings.

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

特許調査や明細書作成の自動化を促進するClaude Fable 5.1

6/9/2026

0 Comments

 
Anthropic社のAIモデルが「Claude Fable 5」から「Claude Fable 5.1」へ進化し、長時間稼働する自律型エージェント能力や科学研究支援が大幅に強化され、特許調査や明細書作成の自動化を促進するとされています。
この「Claude Fable 5.1」の知財業務へ及ぼす影響を生成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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Claude Fable 5.1: Accelerating the Automation of Patent Searches and Patent Specification Drafting
Anthropic’s upgrade from Claude Fable 5 to Claude Fable 5.1 is said to bring substantial improvements in long-running autonomous agent capabilities and scientific research support, helping to accelerate the automation of patent searches and patent specification drafting.
I have asked generative AI to conduct an in-depth analysis of Claude Fable 5.1’s impact on intellectual property (IP) work and invite you to review the findings. Please note, however, that the AI-generated research and analysis are based solely on publicly available information, may not necessarily reflect actual conditions, and may contain inaccuracies. Please keep these limitations in mind when reviewing the results.

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

中国で始まったAIエージェント経済

6/9/2026

0 Comments

 
中国では、AIがユーザーの意図を解釈し取引完了まで代行する「委任型消費(エージェント型コマース)」が日常のインフラになりつつあります。
中国ではスーパーアプリや即時決済網が統合されているため、AIが商取引の直接の執行主体として日常に浸透し、「ひとり会社」の爆発的増加など産業構造を激変させています。対照的に日本は、法整備の遅れやシステムの分断、リスク回避傾向により、AIの活用が情報提示や対話支援の域に留まっています。
この日中の比較を生成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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
中国で始まったAIエージェント経済の衝撃
9/3
https://news.yahoo.co.jp/articles/ff4a7234c3c233fb90cb09c7a842a8c7f934f46c
 
 
The Emergence of the AI Agent Economy in China
In China, “delegated consumption” (agentic commerce)—in which AI interprets users’ intentions and acts on their behalf through to the completion of transactions—is becoming part of the infrastructure of everyday life.
With super apps and instant payment networks integrated, AI is becoming embedded in daily life as a direct executor of commercial transactions, dramatically reshaping the industrial landscape through developments such as the explosive growth of “one-person companies.” By contrast, in Japan, delays in developing legal frameworks, fragmented systems, and a tendency toward risk aversion have kept AI use confined to providing information and conversational support.
I asked generative AI to explore this comparison between China and Japan in greater depth. Please refer to the resulting research and analysis, bearing in mind that the findings are based solely on publicly available information, may not necessarily reflect actual conditions, and may contain inaccuracies.

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が「GPT-6 Astra」を発表

5/9/2026

0 Comments

 
OpenAIが2026年9月3日(米国時間)、GPT-6 Astraを発表しました。初日の時点で使えるのは限られた組織のみで、ChatGPTのPlus/Pro/Business/Enterpriseには「数日以内」に順次提供されるとのことでしたが、9月5日朝の段階で使える状態になっていました。今日は、夕方から移動時間があるので、移動時間に試したいと思います。
「ARC-AGI-3」では、Standard 62.7%、 Provider Adapter 99.9%と、これまでの水準を大きく超えるなど、Computer Use、Browsing、Software Engineering、Science、Professional Workなどで高い性能を持ち、コード、ブラウザ、専門ソフトをまたぐ複数ステップの作業を得意とすると説明しています。
しかし、「Artificial Analysis Intelligence Index」では61で、「Claude Fable 5.1」(66)、「Claude Fable 5」(62)、「Muse Spark 1.3」(62)を下回っています。
特定の機能だけ強化して、その他はあまり変わらず、といったことでしょうか。
「GPT-6 Astra」について生成AIに深掘り調査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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
A new generation of intelligence
https://openai.com/index/gpt-6-astra/
 
 
OpenAI Announces “GPT-6 Astra”
On September 3, 2026 (U.S. time), OpenAI announced GPT-6 Astra. Access on the first day was limited to a small number of organizations, and OpenAI said it would roll out to ChatGPT Plus, Pro, Business, and Enterprise users “within a few days.” By the morning of September 5, however, it was already available. I’ll be traveling this evening, so I plan to try it out while in transit.
On ARC-AGI-3, GPT-6 Astra achieved scores of 62.7% for Standard and 99.9% for Provider Adapter, far surpassing previous performance levels. OpenAI describes the model as delivering strong performance in areas such as Computer Use, Browsing, Software Engineering, Science, and Professional Work, and as excelling at multistep tasks that span code, browsers, and specialized software.
However, its score on the Artificial Analysis Intelligence Index is 61, below Claude Fable 5.1 (66), Claude Fable 5 (62), and Muse Spark 1.3 (62).
Does this mean that only certain capabilities have been enhanced, while the others remain largely unchanged?
I asked generative AI to conduct an in-depth investigation into GPT-6 Astra and invite you to review the findings. Please note, however, that the AI-generated research and analysis are based solely on publicly available information, may not necessarily reflect the actual situation, and could contain inaccuracies. Please keep these limitations in mind when reviewing the results.

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

Claude_Fable_5.1はサイエンスエージェント化したのか

5/9/2026

0 Comments

 
2026年9月1日、Anthropicは最新モデル「Claude Fable 5.1」と「Claude Mythos 5.1」を発表しました。両モデルは、コーディング、AIエージェント、科学研究などの高度なタスクを対象としたモデルで、AIモデルの性能を測定する独立評価プラットフォーム「Artificial Analysis」が提供する、AIモデルの総合的な知能を評価するための複合指標(ベンチマーク)「Artificial Analysis Intelligence Index」では、66で最高評価です。(Claude Opus 5 (max) 63, GPT-5.6 Sol (max) 61)
「コーディングと知識労働のための世界最先端のモデルであり、その研究能力は、AIモデルが科学の進歩にどのように貢献していくかを示す初期段階の展望を示しています。」ということで、Claude_Fable_5.1は「サイエンスエージェント化した」という議論もあります。
Claude_Fable_5.1はサイエンスエージェント化したのか、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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
 
Has Claude Fable 5.1 Become a Science Agent?
On September 1, 2026, Anthropic announced its latest models, Claude Fable 5.1 and Claude Mythos 5.1. Both are designed for advanced tasks, including coding, AI agent workflows, and scientific research. They achieved the highest score of 66 on the Artificial Analysis Intelligence Index, a composite benchmark for evaluating the overall intelligence of AI models, provided by the independent AI evaluation platform Artificial Analysis. For comparison, Claude Opus 5 (max) scored 63, while GPT-5.6 Sol (max) scored 61.
The models are described as “the world’s most advanced models for coding and knowledge work, with research capabilities that offer an early glimpse of how AI models will contribute to scientific progress.” This has prompted some to argue that Claude Fable 5.1 has evolved into a “science agent.”
I asked generative AI to explore in depth whether Claude Fable 5.1 has indeed become a science agent. Please review the findings with the understanding that the AI-generated research and analysis are based solely on publicly available information, may not necessarily reflect the actual state of affairs, and may contain 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.
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サムライによる特許網自動生成システム

4/9/2026

0 Comments

 
日経新聞「トヨタ系のAIサムライ、AIが特許網を生成 中小も技術保護しやすく」によれば、AIサムライは、大阪大学大学院情報科学研究科の鬼塚真教授の研究室との共同研究で出願を予定する基本特許に関連する特許を人工知能(AI)が提案し、特許網を自動生成するサービスを開発しました。新サービスは利用者が基本特許の内容を入力し、構築したい特許網の広さを決めるとAIが5分程度で周辺特許案を5〜6件提案します。また、特許網自動生成システムと外部のAIと接続する「MCP」の提供によって、現在年間360万円程度の利用料金も年60万〜120万円にすることで、導入社数を1年後に300社に増やすという積極策を発表しています。
このAIサムライの動きについて生成AIに深掘り調査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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
トヨタ系のAIサムライ、AIが特許網を生成 中小も技術保護しやすく
2026年9月3日
https://www.nikkei.com/article/DGXZQOUC26BH50W6A820C2000000/
 
 
AI Samurai’s Automated Patent Portfolio Generation System
According to the Nikkei article, “Toyota-Affiliated AI Samurai Uses AI to Generate Patent Portfolios, Making It Easier for SMEs to Protect Their Technologies,” AI Samurai has developed a service that uses artificial intelligence to propose related patent concepts and automatically build a patent portfolio around a core patent that a user plans to file. The service was developed through joint research with the laboratory of Professor Makoto Onizuka at Osaka University’s Graduate School of Information Science and Technology.
With the new service, users enter the details of a core patent and specify how broad they want the surrounding patent portfolio to be. The AI then proposes five or six related patent concepts in approximately five minutes. AI Samurai has also announced an aggressive growth strategy involving the provision of an “MCP” that connects its automated patent portfolio generation system with external AI services. By reducing the current annual fee of approximately ¥3.6 million to between ¥600,000 and ¥1.2 million, the company aims to increase the number of corporate users to 300 within one year.
I asked generative AI to conduct an in-depth investigation into AI Samurai’s latest initiative. Please refer to the findings for further details. Please note, however, that the research and analysis produced 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

Metaが「Muse Spark 1.3」を公開

4/9/2026

0 Comments

 
Metaは9月2日(現地時間)、 「Muse Spark 1.3」 を公開しました。8月初めにリリースされた「Muse Spark 1.2」に続くAIモデルで、エージェント的なタスクとコーディングの性能が向上しているということです。「Artificial Analysis Intelligence Index」では、「max」推論レベルでは63で、Anthropicの「Claude Fable 5」(63)と同じ、OpenAIの「GPT-5.6 Sol」(62)を上回るスコアを記録しており、2026年9月1日に発表された「Fable 5.1」(66)に及ばないもののトップレベルの性能を備えていることがわかります。
「Muse Spark 1.3」について生成AIに深掘り調査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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Muse Spark 1.3
https://developer.meta.com/ai/models/muse-spark/
 
 
Meta Releases “Muse Spark 1.3”
Meta released Muse Spark 1.3 on September 2 (local time). Following Muse Spark 1.2, which was released in early August, the new AI model reportedly delivers improved performance in agentic tasks and coding.
On the Artificial Analysis Intelligence Index, Muse Spark 1.3 achieved a score of 63 at the “max” reasoning level. This matches Anthropic’s Claude Fable 5 at 63 and exceeds OpenAI’s GPT-5.6 Sol at 62. Although it falls short of Fable 5.1, which was announced on September 1, 2026, and scored 66, the results indicate that Muse Spark 1.3 offers top-tier performance.
I asked a generative AI system to conduct an in-depth investigation into Muse Spark 1.3. Please refer to the findings below. Please note, however, that the research and analysis generated by AI are based solely on publicly available information, may not necessarily reflect the full reality, and may 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

Gemini 3.8 Flash発表

3/9/2026

0 Comments

 
Googleは2026年9月2日、新しいGeminiモデル「Gemini 3.8 Flash」と「Gemini 3.8 Flash Cyber」を発表しました。「Gemini 3.8 Flash」は、コーディング、AIエージェント、複雑な推論を強化した汎用モデルです。一方の「Gemini 3.8 Flash Cyber」は、脆弱性の発見や修正など、サイバーセキュリティに特化したモデルです。3週間前にリリースしたFlash 3.7の勢いをそのままに、わずか6週間で3回目のFlashリリースとなります。
AIモデルの性能を測定する独立評価プラットフォーム「Artificial Analysis」が提供する、AIモデルの総合的な知能を評価するための複合指標(ベンチマーク)「Artificial Analysis Intelligence Index」では、「Gemini 3.8 Flash」は59で、昨日リリースされたClaude Fable 5.1が66で最高、Claude Opus 5 (max) 63, GPT-5.6 Sol (max) 61、最高水準には達していないようです。
「Gemini 3.8 Flash」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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/
 
 
Gemini 3.8 Flash
On September 2, 2026, Google announced two new Gemini models: Gemini 3.8 Flash and Gemini 3.8 Flash Cyber. Gemini 3.8 Flash is a general-purpose model with enhanced capabilities in coding, AI agents, and complex reasoning. Gemini 3.8 Flash Cyber, meanwhile, is a specialized model designed for cybersecurity tasks such as identifying and remediating vulnerabilities. Building on the momentum of Flash 3.7, which was released three weeks earlier, this marks the third Flash release in just six weeks.
According to the Artificial Analysis Intelligence Index, a composite benchmark provided by Artificial Analysis, an independent platform for evaluating AI model performance, Gemini 3.8 Flash scored 59. This appears to place it below the very top tier, with Claude Fable 5.1, released yesterday, leading with a score of 66, followed by Claude Opus 5 (max) at 63 and GPT-5.6 Sol (max) at 61.
I asked generative AI to conduct an in-depth analysis of Gemini 3.8 Flash. Please see the results below. 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 situation. 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

オムロン、知財AIエージェントを内製し特許関連工数を50%削減

3/9/2026

0 Comments

 
2026年9月2日付の日経クロステックの記事「オムロンがBedrock活用で知財AIエージェント内製、特許関連工数を50%減」によれば、オムロンは、AWSの生成AI基盤「Amazon Bedrock」などを活用し、先行技術調査と発明説明書作成を支援する知財AIエージェントを自社開発し、2026年1月頃から研究開発部門で利用を始め、これらの業務に要する工数を、少なく見積もっても50%削減したということです。
オムロンでは、発明考案のプロセスを、①発明の創出、②先行技術調査、③発明説明書の作成、の3段階に整理しています。エンジニアが本来注力したいのは発明の創出ですが、専門的な特許文献の確認や説明書作成に多くの時間を取られていました。
そこで、特許庁から取得した過去の出願データをAmazon S3に格納し、Amazon OpenSearch ServiceとAmazon Bedrockを組み合わせました。AIエージェントは、入力されたキーワードや仕様書を基に関連文献を検索・要約し、類似技術の整理、競合分析、発明説明書の作成まで支援します。
商用データベースやSaaSの利用も検討しましたが、データ量の制約や独自の業務プロセスへの対応を考え、内製を選択しました。また、出願前の機密情報を安全に扱うため、AWS上に共通AI基盤「RDinX」を構築し、セキュリティー設定やAIエージェントの挙動、プロンプトを一元的に管理しています。
今回の事例は、生成AIの活用が単なる文書要約から、発明発掘、調査、分析、説明書作成をつなぐ業務プロセスの再設計へ進んでいることを示しています。AIに判断を任せ切るのではなく
、人が結果を確認しながら、エンジニアを定型作業から解放し、発明創出に集中させる取り組みとして注目されます。
この記事の内容を生成AIに深掘り調査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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
オムロンがBedrock活用で知財AIエージェント内製、特許関連工数を50%減
2026.09.02
https://active.nikkeibp.co.jp/atcl/act/19/00798/080300002/
 
 
OMRON Develops In-House IP AI Agent, Cutting Patent-Related Workload by 50%
According to a Nikkei XTECH article published on September 2, 2026, titled “OMRON Develops In-House IP AI Agent Using Bedrock, Cutting Patent-Related Workload by 50%,” OMRON has developed its own intellectual property AI agent using Amazon Web Services’ generative AI platform, Amazon Bedrock, and other AWS services. The agent supports prior-art searches and the preparation of invention disclosure documents. OMRON began using it within its research and development divisions around January 2026 and has conservatively estimated that it has reduced the time required for these tasks by at least 50%.
OMRON divides the invention development process into three stages: (1) generating an invention, (2) conducting a prior-art search, and (3) preparing an invention disclosure document. Although engineers would ideally devote most of their attention to creating inventions, they had been spending considerable time reviewing highly technical patent documents and preparing disclosure documents.
To address this issue, OMRON stored historical patent application data obtained from the Japan Patent Office in Amazon S3 and combined Amazon OpenSearch Service with Amazon Bedrock. Based on keywords or technical specifications entered by users, the AI agent searches for and summarizes relevant documents. It also assists with organizing similar technologies, analyzing competitors, and preparing invention disclosure documents.
OMRON considered using commercial databases and software-as-a-service solutions. However, it chose to develop the system in-house because of restrictions on data volume and the need to accommodate its proprietary business processes. To securely handle confidential information before patent applications are filed, the company also built a shared AI platform called “RDinX” on AWS. The platform centrally manages security settings, AI agent behavior, and prompts.
This case illustrates how the use of generative AI is evolving beyond simple document summarization toward the redesign of entire workflows connecting invention discovery, prior-art research, analysis, and the preparation of disclosure documents. Rather than leaving decisions entirely to AI, OMRON has adopted a human-in-the-loop approach in which people review the AI-generated results. The initiative is particularly noteworthy because it frees engineers from routine tasks and enables them to concentrate more fully on creating inventions.
I asked generative AI to conduct an in-depth investigation of the matters discussed in this article. Please refer to the results below. Please note, however, that the research and analysis generated by AI are based solely on publicly available information. They may not necessarily reflect the actual circumstances and may 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

タイ、英国、韓国、中国、日本のソブリンAI

3/9/2026

0 Comments

 
世界各国の人工知能(AI)戦略が民間主導の技術開発から、国家が主導する社会実装へと移行しています。生成AIに、タイ、英国、韓国、中国、日本の5カ国を対象に調査させましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Sovereign AI in Thailand, the United Kingdom, South Korea, China, and Japan
Artificial intelligence (AI) strategies around the world are shifting from private-sector-led technology development toward government-led implementation across society. I asked generative AI to examine the initiatives of five countries—Thailand, the United Kingdom, South Korea, China, and Japan—and have shared the results for your reference.
Please note that the research and analysis conducted by generative AI are based solely on publicly available information. They may not necessarily reflect the actual situation and may contain inaccuracies.

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

「Claude Fable 5.1」と「Claude Mythos 5.1」発表

2/9/2026

0 Comments

 
2026年9月1日、Anthropicは最新モデル「Claude Fable 5.1」と「Claude Mythos 5.1」を発表しました。両モデルは、コーディング、AIエージェント、科学研究などの高度なタスクを対象としたモデルです。大きなポイントは、Fable 5.1とMythos 5.1が完全に別のモデルではなく、同じ基盤モデルを異なるセーフガードで提供していることです。
AIモデルの性能を測定する独立評価プラットフォーム「Artificial Analysis」が提供する、AIモデルの総合的な知能を評価するための複合指標(ベンチマーク)「Artificial Analysis Intelligence Index」では、66で最高評価です。(Claude Opus 5 (max) 63, GPT-5.6 Sol (max) 61)
「コーディングと知識労働のための世界最先端のモデルであり、その研究能力は、AIモデルが科学の進歩にどのように貢献していくかを示す初期段階の展望を示しています。」ということですので、楽しみです。
Claude Fable 5.1 と Claude Mythos 5.1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、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。
 
Claude: Fable 5.1 and Mythos 5.1
https://www.anthropic.com/claude-fable-and-mythos-5-1
 
 
Anthropic Announces “Claude Fable 5.1” and “Claude Mythos 5.1”
On September 1, 2026, Anthropic announced its latest models, Claude Fable 5.1 and Claude Mythos 5.1. Both models are designed for advanced tasks such as coding, AI agent operations, and scientific research. One of the most notable aspects of the announcement is that Fable 5.1 and Mythos 5.1 are not entirely separate models; rather, they are based on the same underlying foundation model but are offered with different safeguards.
On the Artificial Analysis Intelligence Index—a composite benchmark developed by Artificial Analysis, an independent AI model evaluation platform, to assess the overall intelligence of AI models—the models received the highest score of 66. By comparison, Claude Opus 5 (max) scored 63, while GPT-5.6 Sol (max) scored 61.
Anthropic describes the model as “the world’s most advanced model for coding and knowledge work,” adding that its research capabilities offer an early glimpse of how AI models may contribute to scientific progress. This is certainly something to look forward to.
I asked generative AI to conduct an in-depth analysis of Claude Fable 5.1 and Claude Mythos 5.1, and I invite you to read the results. Please note that this research and analysis were generated by AI based solely on publicly available information. They may not necessarily reflect the actual situation and may 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
<<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