ショートニュース・サーベイ261114

(2026. 10.2~11.14)

  • AIエージェントは制御可能か?
Can AI Agents Be Contained? NVIDIA Moves AI Agent Security Outside the Model
AI agents have always needed limits. The harder problem is making those limits hold when an increasingly capable agent can search for another way to complete it...
  • OpenAI、GPT-6.1 Astraのリリースを中止。安全性への懸念から。
wsj.com
  • 人間どうしのオンライン会議に混ざるAI。

ainews.com/p/tavus-griffin-the-human-like-ai-that-48-of-people-thought-was-real

  • 「従業員の処遇はAIだけで決めるな」ーカリフォルニア州法。
Just a moment...
  • AIで開発されたAIアクセラレーター
GitHub - FeSens/openTPU: An open-source AI accelerator, developed by AI: RTL, ISA, simulator, compiler and profiler in one repo. Runs Qwen3, LFM2.5 and Qwen3.5 on a Kintex-7 PCIe card.
An open-source AI accelerator, developed by AI: RTL, ISA, simulator, compiler and profiler in one repo. Runs Qwen3, LFM2.5 and Qwen3.5 on a Kintex-7 PCIe card. ...
  • ニキビ処方はAIで。
Nolla Derm AI Prescription: Now Piloting in Utah
Nolla Derm is the first AI to issue an initial acne prescription, not just refills. Now piloting in Utah, starting at .99/month. Download the app today.
  • SpaceX、D2Dの通信許可。
FCC Grants SpaceX Approval for 15,000 Direct-to-Device Satellites in MSS Spectrum
The FCC approved SpaceX’s request to deploy 15,000 satellites for direct-to-device (D2D) services, while deferring some of SpaceX’s requests.  This
  • SpaceX、SpaceMobileに向けて。
Checking connection
  • アメリカン航空、全路線でStarlink導入
Checking connection
  • Amazon Leo、衛星の製造1,000機を超える。
Checking connection
  • クロード・ハイク 5.5

https://www.anthropic.com/ claude-haiku-5-5

  • GPT-6、無料ユーザーにも
https://openai.com/ja-JP/index/gpt-6-for-everyone/
  • ジェミニ アルゴン4
Gemini 4 Argon: our next era of frontier intelligence
Announcing Gemini 4 Argon, our frontier model for real-world coding, enterprise knowledge work, and cyber defense, rolling out soon.
  • OpenAI、難解な数学問題の解法に疑問。撤回も。

https://www.chosunonline.com/ site/data/html_dir/2026/10/10/ 2026101080019.html

  • ロシアのLEO
Checking connection
  • バングラデシュでStarlinkMobile
Checking connection
  • 小型のIoT機器複数でAIを動かすことをAIが開発。

“Anthropic、自己複製AIの「兆し」を指摘 55ドルのTVボックス15台で26Bモデルが稼働” https://l.smartnews.com/m-8FPhJjeU/57JDAo

  • 人間の労働を危険に晒す?Amazonでのロボットと人間。
Amazon has over 1 million robot workers. What’s next for human employees?
The second-largest corporate employer has bet big on robot workers to scale.
  • AIでAIをトレーニングする、という「盗難」を99%回避。
Strengthening our defenses against distillation attacks
Illicit distillation allows model developers to copy frontier AI capabilities without creating proportionally robust safeguards. Here, we describe how we’ve red...
  • AMEX、AIが勝手に購入するのをAIが阻止。
Amex introduces a framework to prevent AI agents from making unauthorized purchases | TechCrunch
American Express is introducing a new framework designed to ensure AI agents only make purchases their users have authorized, potentially preventing costly mist...
  • Microsoft、MAI-Base-2、2.8兆パラメータの学習完了。
Frontier Pre-training at Microsoft AI | Microsoft AI
  • Meta、Muse Spark OpenでローカルAI。

ainews.com/p/meta-releases-muse-spark-open-can-businesses-run-powerful-ai-without-the-cloud

  • 次世代の家電はAI入り。
Cerebras CEO says we're nearing a future where everyone has a personal AI assistant at home | TechCrunch
Cerebras CEO Andrew Feldman predicts a future where personal AI assistants become as commonplace as household appliances, running partly on local devices and pa...
  • Anthropic、外部の機関に内部アクセス権限。
Anthropic update on independent evaluators
Anthropic announces updates to embedding independent evaluators
  • Bank of America、AIでの脆弱性検出と対策に、1000人を雇用。
Bank of America hired 1,000 contractors to patch software flaws exposed by Anthropic's Mythos
Bank of America hired 1,000 contractors to patch software flaws exposed by Anthropic’s Mythos, offering a rare look at costs involved in defending against AI.
  • Claude、量子回路を設計。
Designing more efficient quantum circuits with Claude
Working largely on its own, Claude designed quantum circuits that improve on the best published versions, in some cases by an order of magnitude or more.
  • Anthropic社内ではAIが人間の6倍の仕事をしている。
How employment is changing at Anthropic
How has AI restructured work and employment at Anthropic? As a company at the frontier of AI adoption, Anthropic could be a leading indicator of how AI will shi...
  • 従来のソフトウエア開発の仕組みの大きな変革。
Tokenmaxxing and the Collapse of Coding Costs
Two developers used AI coding agents to recreate Claude Code's core architecture in a matter of hours, challenging the economics of traditional software develop...
  • 暴走AIエージェント監視。Microsoft。
Microsoft unveils tools to monitor and corral rogue AI agents
New features in Microsoft 365 E7 and Project Perception will assess each agent's risk, with the ability to block it from making changes, require human intervent...
MSN
  • 「中国のAI」「米国のAI」は同じもの。フォーブス記事。

“「米国のAI」も「中国のAI」も存在しない” https://l.smartnews.com/m-8ItVjqrm/WV7KUJ

  • 米国連邦政府、自組織内のDeepSeekの削除を命令。
US orders federal agencies to remove China’s DeepSeek AI from their systems | TechCrunch
The U.S. has ordered federal civilian agencies to remove DeepSeek products and third-party services built using its technology, citing national security concern...
  • Google CloudでClaudeが利用可能に。
reuters.com
  • AIがphpの脆弱性を発見し公開。世界中が大騒ぎに。
MSN
  • SpaceXの成長はStarlinkに代わってAI主導。
Checking connection
  • オーストラリア、AIの安全性について法律で規制。
Australia could adapt airline and bank laws for AI safety
The federal government is looking to banking and aviation regulation as it develops laws that would make AI companies demonstrate they can identify and manage r...
  • Gemini 4 Argon、Google Claudeで利用可能に。
Gemini 4 Argon available on Google Cloud | Google Cloud Blog
Starting today, we are making our new frontier model, Gemini 4 Argon, available for everyone on Google Cloud.
  • ドイツ軍、対ロシアを考え宇宙政策強化を考える。
Checking connection
  • 静止高軌道、低軌道の「LEO」そして中軌道「MEO」それぞれの役割。
Insight Paper: When Does Orbit Matter | SES
Discover how orbit and network architecture shape performance outcomes. Download the SES insight paper to explore GEO, MEO, and LEO trade-offs.
  • AIがすい臓がんの兆候を1年前に発見。
AI detects “invisible” signs of pancreatic cancer years before diagnosis
A powerful new AI system could help doctors detect pancreatic cancer years before it becomes visible on conventional scans. The technology, called REDMOD, ident...
  • 英国のLCC、Starlinkを採用。
Checking connection
  • Qwenで北京のサーバーにデータ送信を試みている。

Concerning behavior from a popular open-weight AI model | Cohere https://cohere.com/blog/concerning-behaviour-popular-open-weight-ai-model

  • Microsoftの医療診断AI、成功率83%。
Microsoft's Diagnostic AI Scores 83%. Can It Improve Patient Care?
Microsoft's AI is outperforming doctors on diagnostic tests. Could it eventually bring advanced medical expertise to more patients?
  • AIエージェント、銀行口座アクセス。
Just a moment...
  • カナダのTelestar、D2Dサービス事業。
Checking connection
Gemini 4 Carbon: our next frontier model
Gemini 4 Carbon is our strongest model with state-of-the-art cybersecurity and long horizon capabilities.
  • OpenAI、選挙対策センター設置。
Just a moment...
  • Google、宇宙AIデータセンターへ一歩。
We’ve received our first response from the Project Suncatcher prototype satellite.
This week, we reached our "Hello World" milestone: the first-ever AI response from a data center class TPU in orbit.
  • AIがニュースから天変地異を予測。
Google Genie 4: Can AI-Generated Worlds Train Real-World AI?
What if robots could practice handling disasters, navigating obstacles, and making decisions in AI-generated worlds before facing those challenges in real life?
  • 進む「AIの低価格化」。
Claude (@claudeai) on X
We’re renaming our Max plans: Max 5x → Max Max 20x → Ultra Max is still 0/month, and Ultra is still 0. If you’re a subscriber, there’s nothing you need...
  • United航空、2027年末までに全機にStarlinkによる機内Wi-Fiサービス導入。それに伴いセキュリティにはAI活用。
Checking connection
  • AI停止は簡単ではない。
CrowdStrike’s AI Kill Switch: Can It Really Stop AI Agents?
Stopping an AI agent is one thing. Stopping everything it has already set in motion is another—and that's where the real challenge begins.
  • まもなくAI自身がAIを作り出す。
Anthropic warns that it's very close to RSI capability, aka self-building AI | TechCrunch
RSI is the point at which AI grows so good at training itself that it spawns a dramatic acceleration of AI capabilities. To the AI doomers, RSI represents the c...
  • Claudeが不審な支出を検知。
Introducing Personal Finance in Claude
Ask Claude about your money and get straight answers from the accounts you connect.
  • Museでビデオチャット。
Meta says US users can now video chat with their Muse AI avatar | TechCrunch
Meta's AI agent can now take calls as a customizable animated avatar, ahead of the launch of a Tamagotchi-like wearable designed to bring Muse to life.
  • ai上司。
https://openai.com/ja-JP/index/the-future-of-the-firm/?utm_source=www.ainews.com
  • セキュリティは、もはやAIなしでは危険。
https://openai.com/ja-JP/index/introducing-gpt-6-cyber/
  • アプリを開かずに。
Apple's Siri Expansion Could Change What Makes Business Apps Valuable
What happens to the value of a business app when customers can use its services without ever opening it?
  • NVIDIAジェンセン・ファン氏の話題。
Jensen Huang’s affair with alleged Chinese spy: What it means for Nvidia and global AI race
What are the national security implications of the scandal involving Nvidia CEO Jensen Huang? How might it affect Nvidia's future? Could it lead to further chip...
  • AirBnB、旅行計画AIエージェント導入予定。
Airbnb says it's launching an AI travel agent in the first half of 2027 | TechCrunch
Airbnb says its upcoming AI agent will help travelers plan trips, compare options, manage bookings, and get assistance throughout their stay.
  • 宇宙インターネットシステムの脆弱性の元はAPI
Checking connection
  • AIによる支配は3年以内に始まる?
AI Takeover Within Three Years: Paul Christiano's Dire Warning
When “Get a Better Score” Becomes “Get Around the Humans”: AI Researcher Scares a Tech Journalist at the 2026 Hinton Lectures.

コメント

タイトルとURLをコピーしました