|
2026年6月26日、令和7年度 特許庁産業財産権制度問題調査研究報告書が公表されました。テーマは10個で、そのひとつである「AI発明の保護に関する調査研究報告書」(三菱UFJリサーチ&コンサルティング)は、AIを利活用する企業へのヒアリング等を通じて、将来的な発明の保護の在り方に関する課題・対応策等を分析し、AI技術の進歩を踏まえた発明の保護の在り方を検討するための基礎資料とすることを目的として、実施されました。 この報告書について、生成AIに深掘りさせました。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。 特許庁産業財産権制度問題調査研究について https://www.jpo.go.jp/resources/report/sonota/zaisanken-seidomondai.html (9)AI発明の保護に関する調査研究 https://www.jpo.go.jp/resources/report/sonota/document/zaisanken-seidomondai/2025_09_zentai.pdf Survey Research Report on the Protection of AI Inventions On June 26, 2026, the Japan Patent Office published its FY2025 Survey Research Reports on Issues Relating to the Industrial Property Rights System. The reports cover ten themes. One of them, the “Survey Research Report on the Protection of AI Inventions” by Mitsubishi UFJ Research and Consulting, was conducted with the aim of providing foundational materials for considering how inventions should be protected in light of advances in AI technology. The report analyzes issues, possible responses, and related matters concerning the future protection of inventions, based on interviews with companies that use and apply AI. I had generative AI take a deeper look at this report. 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 situation. They may also contain incorrect information, so please keep this in mind when referring to them. Your browser does not 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. Your browser does not 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
オムロンは6月26日のAWSサミット JapanでAmazonベッドロックを 活用した自社開発のAIエージェントを発表しました。このシステムが担うのは知財業務で、特許出願前の先行技術調査や発明説明書の作成といった専門知識が必要な作業をAIが自動で処理し、知財業務の工数を90%削減したと発表、大幅な時間短縮を実現しており、現在はR&D部門を中心に利用者を拡大中とのことです。 オムロンが構築した独自の生成AI基盤「RD Buddy」と、その上で稼働する「知財AIエージェント」の革新的な活用事例となっており、特許情報の膨大化や機密保持といった課題に対し、AWSを活用したセキュアなサーバーレス環境を構築することで、高度な専門業務の自動化を実現しました。RAG(検索拡張生成)技術や自然言語処理を用いることで、複雑な特許分析の工数を劇的に削減し、組織全体の生産性を向上させています。また、単なるツールの導入に留まらず、全社推進プロジェクト「AIZAQ」を通じて、AIを使いこなすための組織文化の醸成やガバナンス体制の確立を並行して進めた点に特徴があります。これらの社内実績を汎用化したサービス「koto-Buddy」として外販するなど、知財活動を起点としたビジネス変革が示されています。 このオムロンの取組について、生成AIに深掘りさせました。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。 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 2026.06.26 https://xtech.nikkei.com/atcl/nxt/column/18/03664/062600003/?utm_source=dlvr.it&utm_medium=mastodon 知財業務を革新するオムロンの知財 AI エージェント実装事例 2025-11-04 https://aws.amazon.com/jp/builders-flash/202511/omron-intellectual-property-ai-agent/ OMRON Develops an In-House IP AI Agent, Reducing Patent-Related Workload by 50% At the AWS Summit Japan on June 26, OMRON unveiled its internally developed AI agent built on Amazon Bedrock. The system is designed to support intellectual property (IP) operations by automatically performing knowledge-intensive tasks such as prior art searches before patent filing and drafting invention disclosure documents. According to OMRON, the AI has reduced the workload associated with these IP operations by as much as 90%, significantly shortening processing time. The company is currently expanding its deployment, primarily within its R&D divisions. This initiative represents an innovative implementation of OMRON's proprietary generative AI platform, "RD Buddy," together with the "IP AI Agent" operating on top of it. To address challenges such as the explosive growth of patent information and stringent confidentiality requirements, OMRON built a secure serverless environment using AWS, enabling the automation of highly specialized IP tasks. By leveraging Retrieval-Augmented Generation (RAG) and natural language processing (NLP), the system dramatically reduces the effort required for complex patent analyses while improving productivity across the organization. Moreover, OMRON's initiative extends beyond simply introducing AI tools. Through its company-wide AI transformation program, "AIZAQ," the company has simultaneously fostered an organizational culture that encourages effective AI utilization while establishing an appropriate governance framework. OMRON has also commercialized the expertise gained through these internal implementations by offering a generalized enterprise AI service called "koto-Buddy." This demonstrates how intellectual property activities can serve as a catalyst for broader business transformation. I asked generative AI to conduct an in-depth analysis of OMRON's initiative. 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 inaccuracies, and should therefore 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. Your browser does not 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. DeepMindは、2026年6月10日、論文「From AGI to ASI」をarXivに公開し、AGIからASIへ至る4つの技術的経路を提示しました。 汎用人工知能(AGI)から人工超知能(ASI)へと至る知能の向上は単一の特異点ではなく、スケーリング、アルゴリズムの刷新、自己改善、マルチエージェント化という4つの経路が相互に作用することで進行すると定義されています。 この件について、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。 From AGI to ASI https://arxiv.org/abs/2606.12683 Four Technological Pathways from AGI to ASI On June 10, 2026, Google DeepMind published the paper From AGI to ASI on arXiv, presenting four technological pathways through which Artificial General Intelligence (AGI) could evolve into Artificial Superintelligence (ASI). According to the paper, the transition from AGI to ASI should not be viewed as a single technological singularity. Rather, it is defined as a process driven by the interaction of four complementary pathways: scaling, algorithmic innovation, self-improvement, and multi-agent systems. I asked generative AI to conduct an in-depth analysis of this topic, and we invite you to read the results below. Please note that the analysis is based solely on publicly available information. As with any AI-generated research, it may not fully 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. Your browser does not 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. 令和8年6月24日に開催された令和8年第8回経済財政諮問会議・第5回日本成長戦略会議の合同会議では、戦略17分野の62項目について2040年度まで累計370兆円超の官民投資ロードマップが提示されました。 日本成長戦略会議における議論について、生成AIに、知的財産戦略の観点から見たときの課題を抽出させましたので、ご参照ください。 経済財政諮問会議(令和8年第8回)・日本成長戦略会議(第5回) https://www.cas.go.jp/jp/seisaku/nipponseichosenryaku/kaigi/dai5/gijishidai.html Challenges in Japan’s Growth Strategy from the Perspective of Intellectual Property Strategy At the joint meeting of the 8th Council on Economic and Fiscal Policy and the 5th Japan Growth Strategy Council, held on June 24, 2026, the Japanese government presented a public–private investment roadmap covering 62 initiatives across 17 strategic sectors, with cumulative investment exceeding ¥370 trillion by FY2040. I asked generative AI to analyze the discussions of the Japan Growth Strategy Council and identify the key challenges from the perspective of intellectual property (IP) strategy. Please refer to the following analysis. Please note that the analysis and observations generated by AI are based solely on publicly available information. They may not fully reflect the actual circumstances and may contain inaccuracies. Readers are advised to exercise appropriate judgment when reviewing the contents. Your browser does not 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. 経済産業省は2026年6月12日、「成長投資ガイダンス」の案を公表し、6月26日までパブリックコメントを募集しています。このガイダンス案は、金融庁と東京証券取引所が策定している「コーポレートガバナンス・コード」の基本的な考え方を前提とし、企業の持続的成長と中長期的な企業価値向上を、成長投資の観点から具体化・実装へとつなげるための任意参照型のガイダンスとなっています。 この「成長投資ガイダンス」案について生成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。 成⻑投資ガイダンス(案) ―価値創造の拡⼤を通じた企業の持続的成⻑に向けて― https://www.meti.go.jp/press/2026/06/20260612006/20260612006-a.pdf 【日本】経産省、「成長投資ガイダンス」案公表。パブコメ募集 https://sustainablejapan.jp/2026/06/15/japan-growth-investment-guidance/126809 Japan's Ministry of Economy, Trade and Industry (METI) Releases the Draft "Growth Investment Guidance" On June 12, 2026, Japan's Ministry of Economy, Trade and Industry (METI) published the draft "Growth Investment Guidance" and invited public comments through June 26, 2026. The draft guidance is based on the fundamental principles of the Corporate Governance Code established by the Financial Services Agency (FSA) and the Tokyo Stock Exchange (TSE). It serves as a voluntary reference framework designed to translate and implement the concepts of sustainable corporate growth and medium- to long-term enhancement of corporate value from the perspective of growth investment. I asked a generative AI system to conduct an in-depth analysis of this draft Growth Investment Guidance, and I hope you will find the results informative. Please note, however, that the AI's research and analysis are based solely on publicly available information. As such, they may not necessarily reflect the actual circumstances, and the results may contain inaccuracies. Please keep these limitations 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. Your browser does not 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. 公正取引委員会・中小企業庁・特許庁は連名で、2026年6月24日、中小企業の知的財産保護に向けたガイドライン「知的財産権・ノウハウ・データの適切 な取引のための優越的地位の濫用等に関する指針」と、その附属資料である「契約書ひな形」を公表しました。 この新たな知的財産取引指針は、大企業による中小企業からの不当な「知財の吸い上げ」を防ぎ、健全な産業構造を構築することを目的としています。この指針は、特許だけでなくノウハウやデータまでも保護対象に含め、独占禁止法における「優越的地位の濫用」に該当する具体的な違反事例を明確に提示しました。あわせて配布された契約書ひな形は、成果物の帰属や対価設定を適正化するための実務的な指針として機能します。さらに、改正された振興基準では、金型の無償保管の禁止や調達担当者の人事評価規定の見直しなど、組織内部の慣習にまで踏み込んだ対策が講じられました。 本件について、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。 (令和8年6月24日)「知的財産権・ノウハウ・データの適切な取引のための優越的地位の濫用等に関する指針」及び「契約書ひな形」の公表について https://www.jftc.go.jp/houdou/pressrelease/2026/jun/260624_chizaitorihiki.html Japan Fair Trade Commission Releases New Guidelines on Intellectual Property Transactions and Model Contract Templates On June 24, 2026, the Japan Fair Trade Commission (JFTC), the Small and Medium Enterprise Agency (SMEA), and the Japan Patent Office (JPO) jointly released the Guidelines on Abuse of a Superior Bargaining Position and Other Issues for Appropriate Transactions of Intellectual Property Rights, Know-how, and Data, together with the accompanying Model Contract Templates, to strengthen the protection of intellectual property held by small and medium-sized enterprises (SMEs). The new IP Transaction Guidelines are intended to prevent the unfair extraction or appropriation of intellectual property by large companies from SMEs and to foster a healthier industrial ecosystem. The Guidelines extend protection beyond patents to include know-how and data, and clearly identify specific examples of conduct that may constitute an abuse of a superior bargaining position under Japan's Antimonopoly Act. The accompanying Model Contract Templates provide practical guidance for ensuring appropriate ownership of deliverables and fair compensation arrangements. In addition, the revised Promotion Standards introduce measures that go beyond contractual practices and address organizational behavior itself, including the prohibition of requiring suppliers to store molds without compensation and the revision of personnel evaluation criteria for procurement staff. I asked generative AI to conduct an in-depth analysis of these developments. Please note, however, that the analysis is based solely on publicly available information and may not necessarily reflect the actual situation. It may also contain inaccuracies or errors, and therefore should be used 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. 2026年6月16日東京で五庁審判部ユーザーシンポジウムが、6月17日仙台で五庁審判部ハイレベル会合が開催されました。五つの特許庁(日本、アメリカ、欧州、中国、韓国)とWIPOの代表者が、AIやデジタル技術を活用した審理の効率化と、知的財産紛争における国際的な連携強化について議論を交わしたということです。 この内容を生成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。 五庁審判部ハイレベル会合を開催しました https://www.jpo.go.jp/news/ugoki/202606/2026062301.html IP5 Trial and Appeal Boards User Symposium and High-Level Meeting The IP5 Trial and Appeal Boards User Symposium was held in Tokyo on June 16, 2026, followed by the IP5 Trial and Appeal Boards High-Level Meeting in Sendai on June 17, 2026. Representatives from the five major intellectual property offices—Japan, the United States, Europe, China, and South Korea—as well as WIPO, exchanged views on enhancing the efficiency of trial and appeal proceedings through the use of AI and digital technologies, and on strengthening international cooperation in intellectual property dispute resolution. I asked generative AI to conduct an in-depth analysis of these developments, and I would like to share the results for your reference. 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 situation. They may also contain inaccuracies or erroneous information. Accordingly, please review the following content 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. 2026年6月12日に東京で開催された第19回五大特許庁(IP5)長官会合において、日米欧中韓の特許庁は、AI分野における新たな協力ロードマップの推進と、実務レベルの「新たな作業部会(ワーキンググループ)」の設立に正式に合意しました。 世界全体の特許出願の約85%を占める五大特許庁(日本・米国・欧州・中国・韓国)による今回の合意は、今後の世界の特許ルールの「デファクトスタンダード(事実上の標準)」となる極めて重要な決定です。 この長官会合の内容について生成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。 第19回五庁長官会合が開催され、AI分野における今後の五庁の協力について議論しました https://www.jpo.go.jp/news/ugoki/202606/2026061603.html The 19th IP5 Heads of Office Meeting At the 19th IP5 Heads of Office Meeting held in Tokyo on June 12, 2026, the patent offices of Japan, the United States, Europe, China, and South Korea formally agreed to promote a new cooperation roadmap in the field of artificial intelligence (AI) and to establish a new working group at the operational level. This agreement among the IP5 Offices—Japan, the United States, Europe, China, and South Korea—which collectively handle approximately 85% of all patent applications filed worldwide, represents an extremely significant development that is likely to become the de facto standard for future global patent practices and rules. I asked generative AI to conduct an in-depth analysis of the outcomes and implications of this Heads of Office Meeting. Please refer to the following analysis for further details. Please note that the research and analysis generated by AI are based solely on publicly available information. They may not necessarily reflect the actual intentions or circumstances of the relevant organizations and may contain inaccuracies or errors. Accordingly, the information should be reviewed and 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. Your browser does not 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. 世界五大特許庁(IP5)は、急増する特許出願に効率的かつ正確に対応するため、AIを審査プロセスの中核へ導入する戦略を加速させており、各国の特許行政は、単なるツールの活用から、AIと人間が共存する次世代の知財エコシステムの構築へと移行しています。 日米欧中韓の各庁のAI活用動向について、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。 Building a Next-Generation Intellectual Property Ecosystem Where AI and Humans Coexist The world’s five largest patent offices (IP5) are accelerating their strategies to integrate AI into the core of the patent examination process in order to respond efficiently and accurately to the rapidly growing volume of patent applications. As a result, patent administration in each jurisdiction is evolving beyond the mere use of AI tools toward the creation of a next-generation intellectual property ecosystem in which AI and human experts work together in a complementary manner. I asked generative AI to investigate the latest AI utilization trends at the patent offices of Japan, the United States, Europe, China, and South Korea, and I would like to share the findings for your reference. Please note, however, that the research and analysis generated by AI are based solely on publicly available information and may not necessarily reflect actual practices. They may also contain inaccuracies or errors. Accordingly, please review the following content 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. Sakana AI が2026年6月22日に「Sakana Fugu」を一般提供開始しました。上位版「Fugu Ultra」が、一部のベンチマークで Anthropic の Fable 5・Mythos Preview に並ぶ性能となっています。料金はサブスク月20〜200ドル、従量課金併用です。 この「Sakana Fugu」は、巨大な単一モデルの構築に頼る従来のアプローチとは異なり、TRINITYやConductorといった技術を用いて外部のモデル群を自律的に指揮し、複雑な課題に対して高い推論能力を発揮します。この特定のプロバイダーに依存しない設計は、地政学的なリスクや輸出規制によるアクセス遮断を回避する手段となり得ます。 生成AIに「Sakana Fugu」について深堀させましたので、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。 Sakana Fugu マルチエージェントを指揮する、一つのモデル https://sakana.ai/fugu/ Sakana AI、一部「ミュトス越えの性能」うたうAIを提供 複数モデルの“集合知”を活用https://news.yahoo.co.jp/articles/435ac162d0c31854dbd0095766c9cab6f575426b Sakana AI’s “Sakana Fugu” On June 22, 2026, Sakana AI made “Sakana Fugu” generally available. Its higher-end version, “Fugu Ultra,” achieves performance comparable to Anthropic’s Fable 5 and Mythos Preview on some benchmarks. Pricing is based on a subscription model ranging from $20 to $200 per month, combined with usage-based billing. Unlike the conventional approach of building a single massive model, “Sakana Fugu” uses technologies such as TRINITY and Conductor to autonomously orchestrate groups of external models, enabling it to demonstrate strong reasoning capabilities on complex tasks. This provider-agnostic design could serve as a way to avoid disruptions in access caused by geopolitical risks or export controls. I asked generative AI to take a deeper look into “Sakana Fugu,” so please refer to the findings. 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 situation. They may also contain inaccuracies, so please keep this in mind when reviewing them. Your browser does not 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. 日本・米国・欧州・中国・韓国の五大特許庁(IP5)では、審査の効率化と品質向上を目的に、先行技術調査や画像検索、文書要約などの実務へAIを導入しており、人間が最終判断を担う「ハイブリッド審査」への移行が共通の潮流となっています。 特に生成AIの適用は実証段階から本格運用への転換点にあり、国際的な連携や法的なガイドラインの整備も急速に進展しています。一方で、AIによる検索精度の向上は出願人にとってのハードルを上げており、より具体的で論理的な明細書作成といった戦略的対応が重要になってきています。 各国の到達度や制度設計には差異があるため、実務家には最新の運用実態を正確に把握することが求められています。 生成AIに、日本・米国・欧州・中国・韓国の五大特許庁(IP5)における特許審査へのAI活用に関する最新動向を調査させ、多角的に整理させました。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 Utilization in Patent Examination Across the IP5 Patent Offices The five major patent offices (IP5)—Japan, the United States, Europe, China, and South Korea—are increasingly adopting AI technologies in patent examination processes, including prior art searches, image retrieval, and document summarization, with the goals of improving examination efficiency and enhancing quality. A common trend across these offices is the transition toward a “hybrid examination” model in which AI supports examiners while final decisions remain the responsibility of human experts. The application of generative AI, in particular, is reaching a turning point, moving from pilot programs and proof-of-concept projects toward full-scale operational deployment. At the same time, international cooperation and the development of legal and procedural guidelines governing AI use are advancing rapidly. On the other hand, improvements in AI-driven search capabilities are raising the bar for patent applicants. As patent offices become more effective at identifying relevant prior art, strategic responses—such as drafting specifications with greater technical specificity, logical consistency, and robust support for inventive step—are becoming increasingly important. Because significant differences remain among countries in terms of implementation maturity, system design, and regulatory frameworks, patent practitioners must maintain an accurate understanding of the latest operational developments in each jurisdiction. I asked a generative AI system to investigate and organize the latest developments regarding the use of AI in patent examination at the IP5 patent offices—Japan, the United States, Europe, China, and South Korea—from multiple perspectives. Please note that the findings and analyses generated by AI are based solely on publicly available information and may not necessarily reflect actual circumstances. They may also contain inaccuracies, and should therefore be reviewed 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. 日本発のAI企業である株式会社Preferred Networks(PFN)は、他社のモデルに依存しないフルスクラッチ開発にこだわり、AI用プロセッサからクラウド基盤、特定産業向けソリューションまでを一貫して手掛ける垂直統合型のビジネスモデルを構築しています。 2026年6月22日商用版がリリースされたPLaMo 3.0 Primeは、単に次の言葉を予測するのではなく、論理的にステップを追って回答を導き出す「推論(Reasoning)」に特化したモデルで、複雑な関数呼び出し(BFCL)や難度の高い英語推論(GPQA)においては依然としてグローバルな最先端モデルに及ばない領域もありますが、日本語を起点とした論理構築においては、優位性を発揮するということです。 生成AIにこの件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。 国産生成AI基盤モデルPLaMo 3.0 Primeを正式リリース https://www.preferred.jp/ja/news/pr20260622 国産LLM Preferred Networks PLaMo 2.0 Prime、PLaMo翻訳、PLaMo API、PLaMo Chat、PLaMo Lite https://www.youtube.com/watch?v=f3ZqQW83Iw8 Preferred Networks (PFN) and PLaMo Preferred Networks (PFN), a Japanese AI company, has built a vertically integrated business model that emphasizes full-scratch development rather than relying on third-party foundation models. The company develops the entire AI stack in-house, ranging from AI processors and cloud infrastructure to industry-specific AI solutions. PLaMo 3.0 Prime, which is scheduled for commercial release in June 22 2026, is a reasoning-focused model designed not merely to predict the next word, but to derive answers through logical, step-by-step reasoning. While there remain areas where it does not yet match the world's leading frontier models—particularly in complex function-calling benchmarks (BFCL) and advanced English reasoning tasks (GPQA)—PFN claims that the model demonstrates a competitive advantage in logical reasoning and knowledge structuring when operating from a Japanese-language foundation. I asked a generative AI system to conduct a deeper analysis of this topic, and the results are presented below for your reference. Please note that the investigation and analysis were conducted solely on the basis of publicly available information. As such, the findings may not necessarily reflect the actual situation and may contain inaccuracies. Readers are therefore encouraged to review 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. 2026年4月24日、味の素の100%子会社で、グループ全体の経理を引き受けている、味の素フィナンシャル・ソリューションズ(味の素FS)が経理の承認業務をAIエージェントに任せ始めたという発表が、経理業界をざわつかせたそうです。このAIエージェントは2026年2月から本番稼働していて、すでに現場で動いているのがポイントです。 味の素グループの取り組みがこれまでの取り組みと違うのは、「最終的な承認の判断」そのものをAIに任せた点で、人が確認していた最後の関所を、AIが自律的にさばきます。大企業の本番業務で、月1万件という規模で動かしている実績があるというのは驚きです。 生成AIにこの件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。 味の素の経理AIで工数76%削減|年1万時間創出 2026-06-20 https://aifriends.jp/ajinomoto-accounting-ai-agent/ Ajinomoto’s Accounting AI Cuts Workload by 76% On April 24, 2026, it was announced that Ajinomoto Financial Solutions (Ajinomoto FS), a wholly owned subsidiary of the Ajinomoto Group responsible for handling the group’s accounting operations, had begun delegating accounting approval tasks to an AI agent. The announcement reportedly generated considerable attention within the accounting industry. Notably, this AI agent had already entered full-scale production operation in February 2026 and is currently being used in actual business processes. What distinguishes the Ajinomoto Group’s initiative from previous efforts is that it entrusts the final approval decision itself to AI. In other words, the last checkpoint that had traditionally required human review is now processed autonomously by the AI agent. It is remarkable that this system is already operating in a large-scale corporate environment, handling approximately 10,000 transactions per month in production. I asked a generative AI system to conduct a deeper analysis of this initiative, and the results are presented below for your reference. Please note that the investigation and analysis were conducted solely on the basis of publicly available information. As such, the findings may not necessarily reflect the actual situation in full and may contain inaccuracies. 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. 「知的財産推進計画2026」は、2026年6月12日、第55回知的財産戦略本部(本部長・高市早苗首相)で決定され、内閣府知的財産戦略推進事務局所管となっており、重点方向性の柱の一つに「生成AI等の新たな時代に即した知的財産の保護」を掲げています。また、海外での知財侵害に対する集団訴訟制度の整備等も盛り込まれています。 「AI基本計画(素案)」は、2026年6月19日、人工知能戦略本部(第4回)がとりまとめ、同日e-Govでパブリックコメントが開始(締切6月23日)されており、内閣府科学技術・イノベーション推進事務局人工知能政策推進室所管となっています。。本素案は2025年12月23日閣議決定の我が国初の「人工知能基本計画」(副題「~『信頼できるAI』による『日本再起』~」)の改定版=第Ⅱ期計画であり、「当面は毎年変更」と明記する年次改定方式をとっています。 知的財産推進計画2026とAI基本計画素案における生成AI施策の整合性に関して、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。 Consistency of Generative AI Policies Between the Intellectual Property Strategic Program 2026 and the Draft AI Basic Plan The Intellectual Property Strategic Program 2026 was approved on June 12, 2026, at the 55th meeting of the Intellectual Property Strategy Headquarters (headed by Prime Minister Sanae Takaichi) and is administered by the Intellectual Property Strategy Promotion Office of the Cabinet Office. One of its key policy pillars is the “protection of intellectual property suited to the new era of generative AI and other emerging technologies.” The program also includes measures such as the development of a collective litigation framework to address intellectual property infringement occurring overseas. The Draft AI Basic Plan was compiled on June 19, 2026, by the Artificial Intelligence Strategy Headquarters (4th meeting), and a public comment period was launched on the same day through e-Gov (closing on June 23). The draft is administered by the AI Policy Promotion Office within the Office for Science, Technology and Innovation Policy of the Cabinet Office. It represents the revised version—Phase II—of Japan’s first AI Basic Plan, officially adopted by the Cabinet on December 23, 2025, under the subtitle “Revitalizing Japan Through Trustworthy AI.” The draft adopts an annual revision framework, explicitly stating that the plan will be updated every year for the foreseeable future. I asked a generative AI system to conduct an in-depth analysis of the consistency between the generative AI-related measures contained in the Intellectual Property Strategic Program 2026 and those proposed in the Draft AI Basic Plan. The results are provided below for reference. Please note that the research and analysis produced by generative AI are based solely on publicly available information and may not necessarily reflect the actual situation. They may also contain inaccuracies or errors. Accordingly, 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. Your browser does not 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. 2026年6月19日、人工知能戦略本部(第4回)が持ち回り開催され、「AI法に基づく基本計画(素案)」が、令和8年6月19日(金)から令和8年6月23日(火)までの間という短い期間のパブリックコメントに付されました。 「AI法に基づく基本計画(素案)」は、AIを社会の基盤として取り入れることで「日本再起」を目指す国家戦略の全容を示したものです。急速に普及する自律行動型AI(Agentic AI)を念頭に、日本独自の強みである現場力を活かした「バーティカルAI」や「フィジカルAI」への重点投資を通じて、産業と行政の構造を根底から変革する「AIトランスフォーメーション(AX)」の断行を掲げています。具体策として、政府専用AI「源内(げんない)」による行政の高度化や、特定の企業に依存しない「AI主権」の確立、さらに国際的な信頼を構築するための「責任あるアジャイル・ガバナンス」の主導が提唱されています。AIを単なるツールではなく協働のパートナーと定義し、リスクを適切に管理しながら人間中心のAI社会を実現し、世界で最もAIを開発・活用しやすい国を目指すことが本計画の究極の目的とされています。 生成AIに、この「AI法に基づく基本計画(素案)」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。 人工知能基本計画(素案)に関する御意見の募集について 令和8年6月19日 内閣府 科学技術・イノベーション推進事務局 人工知能基本計画の改定に向け、人工知能基本計画(素案)について、令和8年6月19日(金)から令和8年6月23日(火)までの間、意見を募集します。 https://www8.cao.go.jp/cstp/stmain/20260619ai.html Draft Basic Plan under the AI Act On June 19, 2026, the fourth meeting of the AI Strategy Headquarters was convened by written circulation, and the “Draft Basic Plan under the AI Act” was released for public comment during the relatively short period from June 19 (Friday) to June 23 (Tuesday), 2026. The Draft Basic Plan presents the Japanese government's comprehensive national strategy for achieving a “Revitalization of Japan” through the integration of artificial intelligence as a foundational element of society. With the rapid emergence of autonomous, action-oriented AI systems (Agentic AI) in mind, the plan advocates a decisive AI Transformation (AX) that will fundamentally reshape both industry and public administration. It emphasizes strategic investment in Vertical AI and Physical AI, areas in which Japan can leverage its traditional strengths in operational excellence and on-site expertise. Specific initiatives include enhancing government operations through the dedicated governmental AI system “Gennai,” establishing AI sovereignty that is not dependent on any particular company or technology provider, and leading the international development of “Responsible Agile Governance” to foster global trust in AI. The plan defines AI not merely as a tool but as a collaborative partner, and seeks to realize a human-centered AI society by appropriately managing risks while maximizing opportunities. Its ultimate objective is to make Japan the most attractive and accessible country in the world for AI development and utilization. I asked a generative AI system to conduct an in-depth analysis of this “Draft Basic Plan under the AI Act.” Please note that the findings and analyses generated by AI are based solely on publicly available information and may not necessarily reflect the actual situation. They may also contain inaccuracies or errors, and should therefore be reviewed 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. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. 2026年6月15日から17日にかけて、フランス東部のリゾート地エヴィアン=レ=バンで開催された主要7カ国首脳会議(G7エヴィアン・サミット)の昼食会には、OpenAIやAnthropic、Google、Metaなど、世界の最先端AI企業の創業者やトップらが招かれました。日本から唯一招待されたAI開発企業「Sakana AI」の共同創業者・伊藤社長への単独インタビューの完全版動画『AIは「国家の安全保障」へ。G7で世界のビッグテックと首脳が共有した"危機感"【サカナAI伊藤社長】』を視聴しました。G7サミットでの議論、AIが単なる技術革新の枠を超えて国家安全保障の核心へと移行した現状が良くわかりました。 生成AIにこの会議(昼食会)での議論を深掘りさせました。なお、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は「国家の安全保障」へ。G7で世界のビッグテックと首脳が共有した"危機感"【サカナAI伊藤社長】 https://www.youtube.com/watch?v=fs4VrUlP38U Sakana AI CEO Ito Invited to the G7 Summit Luncheon During the G7 Evian Summit, held from June 15 to 17, 2026, in the French resort town of Évian-les-Bains, a special luncheon brought together heads of state from the Group of Seven and leaders of the world's most advanced AI companies, including founders and top executives from OpenAI, Anthropic, Google, and Meta. I watched the full version of the exclusive interview with Sakana AI co-founder and CEO, Mr. Lion Ito, titled “AI Is Becoming a Matter of National Security: The Sense of Crisis Shared by Global Big Tech Leaders and Heads of State at the G7”. Sakana AI was the only AI development company from Japan invited to participate in the luncheon. The interview provided valuable insight into the discussions held at the summit and clearly illustrated how AI has moved beyond the realm of technological innovation to become a core issue of national security. I asked a generative AI system to conduct a deeper analysis of the discussions that took place at this luncheon. 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 inaccuracies, and should therefore be referenced 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. ミュンヘン地方裁判所は、2026年5月28日、Googleの「AIによる概要」をGoogle自身に帰属する内容と位置づけ、Googleを直接の責任主体と認めて差止めを命じました。Googleの「AIによる概要」について、AI生成概要をGoogle自身に帰属する内容と明確に位置づけた先駆的判断のようです。 生成AIに、この判決について深堀させました。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。 2026/6/17 [WED] Google「AIによる概要」に責任認める判決 ドイツ裁判所、企業への虚偽表示で差止め https://ledge.ai/articles/google_ai_overviews_germany_court_ruling 2026.06.17 「AIによる概要」の虚偽記述はグーグルの責任──ドイツの裁判所が判断 https://wired.jp/article/a-court-has-ruled-that-google-is-liable-for-false-statements-generated-by-ai-overviews/ Court Ruling Holds Google Responsible for “AI Overviews” On May 28, 2026, the Munich Regional Court ruled that Google's “AI Overviews” should be regarded as content attributable to Google itself and recognized Google as the directly responsible party, issuing an injunction against the company. This appears to be a landmark decision that explicitly characterizes AI-generated summaries in Google Search as Google's own content rather than merely third-party information. I asked a generative AI system to conduct an in-depth analysis of this ruling. Please note that the investigation and analysis were generated solely from publicly available information and may not necessarily reflect the full reality of the situation. The analysis may also contain inaccuracies, and therefore should be reviewed 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. AIの進化により、AIを活用して発明を生み出すことが多くなってきました。発明者は自然人に限るというのがグローバルな原則となっていますが、AIの支援を受けた発明を想定したとき、日本と米国の発明者認定基準の相違が大きく影響するケースが考えられます。米国は人間が具体的な解決策を事前に把握していたかを問う「着想(Conception)」を重視し、日本は発明の核心部分への「創作的寄与」を幅広く評価する傾向にありますので、この解釈の差により、人間が課題設定や事後検証のみを行った場合、米国では権利化が困難になる一方で、日本では維持できるというリスクが生じ得ます。特に創薬や材料探索などAIの自律性が高い分野でこの懸念が大きくなっています。 生成AIに、日本と米国の発明者認定基準の相違が及ぼす影響について深堀させました。問題提起として、今後議論が深まることを期待しています。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 Impact of Differences in Inventorship Standards on AI-Assisted Inventions As AI continues to advance, inventions generated with the assistance of AI are becoming increasingly common. While the global principle remains that inventors must be natural persons, significant issues may arise when AI-assisted inventions are evaluated under the differing inventorship standards of Japan and the United States. The United States places strong emphasis on “conception,” requiring that a human inventor possess a definite and permanent idea of the specific solution before the invention is completed. In contrast, Japan tends to evaluate inventorship more broadly based on the inventor’s “creative contribution” to the core aspects of the invention. Because of this difference in interpretation, situations may arise in which a person merely defines the problem and later verifies the results generated by AI. In such cases, obtaining patent protection may be difficult in the United States, while patent rights could still be maintained in Japan. This divergence is particularly concerning in fields where AI operates with a high degree of autonomy, such as drug discovery, materials exploration, and molecular design. As AI systems increasingly contribute to the generation of novel technical solutions, differences in inventorship standards may lead to significant discrepancies in patent eligibility and enforceability across jurisdictions. I asked generative AI to conduct an in-depth analysis of how these differences between Japanese and U.S. inventorship standards may affect AI-assisted inventions. I hope this serves as a starting point for further discussion and debate on this important issue. 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. Readers are encouraged to review 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. Your browser does not 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. 2026年6月12日、オムロンは、東京都内で会見を開き、同社の研究子会社であるオムロン サイニックエックスが研究開発に取り組むAI技術「Decentralized X(DcX)」の概要について説明しました。 この「Decentralized X(DcX)」は、複数の現場にある生データを一箇所に集めずにAIを賢くする画期的な技術で、従来の手法ではプライバシーや機密保持が壁となりデータの統合が困難でしたが、本技術は各拠点で学習した「知識(モデルの出力)」のみを統合することで、安全に高性能なAIを構築できるということです。 アプリズムとの共同検証では、馬の異状検知AIにおいて開発期間を約75%短縮し、暗所での精度を大幅に向上させる成果を上げています。この「知識蒸留」を用いたアプローチは、異なるAIモデル同士でも連携できる柔軟性を持ち、製造や医療など多岐にわたる分野への応用が期待されています。オムロンは、この基盤技術を通じて自律的に成長する「Agentic AI」の社会実装を加速させようとしています。 について、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の学校”で開発期間を7割削減、“蒸留”するオムロンのAI学習手法「DcX」とは https://monoist.itmedia.co.jp/mn/articles/2606/16/news043.html#utm_medium=email&utm_source=mn-day&utm_campaign=20260617 2026年6月09日 オムロンとアプリズム、非集中学習技術「DcX」で開発効率を向上 環境変化に対応できるAIモデルを短期間で開発可能 https://www.omron.com/jp/ja/news/2026/06/c0609.html OMRON’s Decentralized Learning Technology “Decentralized X (DcX)” On June 12, 2026, OMRON held a press briefing in Tokyo and introduced the outline of its AI technology, “Decentralized X (DcX),” which is being developed by its research subsidiary, OMRON SINIC X. Decentralized X (DcX) is a groundbreaking technology that enables AI systems to become smarter without collecting raw data from multiple sites into a single location. Under conventional approaches, privacy concerns and confidentiality requirements often make data integration difficult. In contrast, DcX safely builds high-performance AI by integrating only the “knowledge” learned at each site (i.e., model outputs) rather than the underlying data itself. In a joint verification project with Aprism, the technology reportedly reduced the development time of an AI system for detecting abnormalities in horses by approximately 75% while significantly improving detection accuracy in low-light environments. This approach, based on knowledge distillation, also offers the flexibility to enable collaboration among different AI models, making it applicable across a wide range of fields, including manufacturing and healthcare. Through this foundational technology, OMRON aims to accelerate the social implementation of Agentic AI—AI systems capable of autonomous growth and adaptation. I asked generative AI to conduct an in-depth investigation of this technology, and I would like to share the results for your reference. Please note that the findings and analyses generated by AI are based solely on publicly available information and may not necessarily reflect the actual situation. They may also contain inaccuracies, so please review them 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. フランス発のAI企業であるMistral AIはGoogle DeepMindやMeta出身のエリート研究者によって設立され、米国の巨大テック企業による知能独占に対抗する「欧州のデジタル主権」の象徴として位置づけられています。 技術面では、計算効率に優れた「混合エキスパート(MoE)」と、モデルの重みを公開する「オープンウェイト」戦略を主軸に、高騰するAI開発コストの抑制と透明性の確保を両立させています。 事業面では、単純なモデル提供から自前データセンターや産業用AIの買収へと拡大し、インフラからアプリまでを垂直統合するフルスタックAI企業への変貌を遂げつつあります。 このMistral AIについて、生成AIに深掘りさせました。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。 欧州発のソブリンAI「Mistral AI」 CEO Arthur Mensch(アルチュール・メンシュ)。Mixture of Expertsとオープン・ウェイトの哲学 https://www.youtube.com/watch?v=Ue2Fu-YYva4 Mistral AI: Europe’s Sovereign AI Champion Mistral AI, a French AI company founded by elite researchers from Google DeepMind and Meta, has emerged as a symbol of Europe’s pursuit of “digital sovereignty” in response to the growing concentration of AI power among major U.S. technology companies. From a technological perspective, Mistral AI has built its strategy around two key pillars: the highly efficient Mixture-of-Experts (MoE) architecture and an open-weight approach that publicly releases model weights. This combination seeks to balance the rising costs of AI development with the need for transparency and accessibility. On the business side, the company has expanded beyond simply providing AI models. Through investments in its own data center infrastructure and acquisitions in the industrial AI sector, Mistral AI is evolving into a full-stack AI company that vertically integrates the entire value chain—from infrastructure to applications. I asked a generative AI system to conduct an in-depth analysis of Mistral AI. Please note that the following research and analysis are based solely on publicly available information generated by AI. They may not fully reflect the actual situation and could contain inaccuracies or errors. Please keep this in mind when reviewing the material. Your browser does not 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. |
著者萬秀憲 アーカイブ
April 2026
カテゴリー |
RSS Feed