（2025.6.20～7.18）
●OpenAIが独自ブラウザをリリース。ターゲットはChromeユーザー。
https://www.ainews.com/p/openai-to-launch-ai-powered-web-browser-taking-direct-aim-at-google-chrome
●300ドル以下の、オープンソースのAIロボット。
●宇宙軍司令部が選んだシステムは….
Space Systems Command Gives Nod to Boeing Over Northrop Grumman for ESS Satellites
Space Force Space Systems Command (SSC) gave a boost to Boeing's confidence last week, as SSC chose the company over Northrop Grumman for a .8 billion contrac...
www.defensedaily.com
●オルガノイド研究とは。
Access Denied
blog.mdpi.jp.wordpress.sciforum.net
●米合衆国政府、中国へのAIチップなどの輸出規制を緩和か。
NVIDIA CEO Jensen Huang Promotes AI in Washington, DC and China
This month, NVIDIA founder and CEO Jensen Huang promoted AI in both Washington, D.C. and Beijing — emphasizing the benefits that AI will bring to business and s...
blogs.nvidia.com
●自然な日本語のおしゃべりをするAI。
J-Moshi
nu-dialogue.github.io
●手術部位の感染をAIで発見。
Access Denied
newsnetwork.mayoclinic.org
●Amazon/Project Kuiper、3度めの打ち上げミッションは、SpaceXで行われる。
Amazon's Third Kuiper Mission Launches With Competitor SpaceX
SpaceX launched the third Amazon Project Kuiper mission early Wednesday morning from Cape Canaveral, marking the first SpaceX launch in Amazon’s
www.satellitetoday.com
●BMWと衛星。
The Roadmap to the Satellite-Connected Car with BMW Group and 5GAA
This week’s episode features two experts in vehicle connectivity from BMW Group and 5GAA on the path to integrating satellite connectivity into vehicles.
www.satellitetoday.com
●AIが創造的知性をもちはじめた？
LLMs Show Strategic Personalities in 140K Prisoner’s Dilemma Tests
Study finds OpenAI, Google, and Anthropic LLMs develop unique strategic behaviors in Prisoner's Dilemma tests, signaling evolving reasoning capabilities.
www.ainews.com
●AI音楽クリエイターも加わる音楽コンテスト。
Future Sound Awards Launches to Celebrate AI Music Creators
Future Sound Awards—the first-of-its-kind AI music competition—launched July 1 with K in prizes. Entries close July 22.
www.ainews.com
●AIが設計した医薬品の臨床試験開始。
Isomorphic Labs Prepares for First Human Trials of AI-Designed Drugs
Alphabet’s Isomorphic Labs nears human trials of AI-designed drugs, aiming to speed up development and improve success rates using AlphaFold
www.ainews.com
●デジタルビームフォーミング技術デモ。
MDA Space Reports Digital Beam-Forming Demonstration With Aurora Technology
MDA Space has reported a digital beam-forming milestone with its technology for the Aurora satellite platform.
www.satellitetoday.com
●AIに店舗経営をさせてみました。
AIエージェントに店舗経営を任せる実験結果をAnthropicが発表、はたしてAIに黒字経営は可能なのか？
AIチャットボットのClaudeを開発するAnthropicと、AIの安全性を専門とする企業であるAndon Labsが、Claude Sonnet 3.7のインスタンスにオフィスの自動販売機を約1カ月間経営させるという実験「Project Vend」を行った結果を公開しています。
gigazine.net
●デンマークは、顔や声などに著作権を設定する。
Denmark Moves to Let Citizens Copyright Their Own Faces, Voices
Denmark is preparing legislation to let individuals copyright their face, voice, and body—a direct response to deepfakes and generative AI misuse.
www.ainews.com
●MicrosoftのAI、医療分野で複雑な事象について85%の確率で正確な回答。
Microsoft AI Diagnoses Complex Medical Cases with 85% Accuracy
Microsoft’s AI MAI-DxO diagnoses NEJM case studies with 85% accuracy, outperforming doctors and reducing diagnostic costs
www.ainews.com
●Amazon、AIをベースとした100万台のロボットで配送業務の劇的な合理化を目指す。
Amazon launches a new AI foundation model to power its robotic fleet and deploys its 1 millionth robot
New AI technology will make the world's largest fleet of industrial mobile robots smarter and more efficient.
www.aboutamazon.com
●LLMは理解したことを応用できない。
AIは「賢いフリ」をしていた──ハーバード大などが暴いたLLMの決定的弱点「ポチョムキン理解」とは？ | XenoSpectrum
大規模言語モデル（LLM）は、流暢な会話をこなし、専門的な質問にも答える。その驚くべき能力に、私たちは「AIは本当に理解しているのではないか」という期待を抱きがちだ。しかし、その知性は本物なのだろうか？
xenospectrum.com
●地球観測マルチスペクトルLEO衛星。
BlackSky to Expand Constellation With Multispectral, Large-Area Satellites
BlackSky is planning to expand the capabilities of its EO constellation with multispectral, large-area collection satellites called AROS.
www.satellitetoday.com
●AIのある未来は、今とは違う。「今とは違う」のを嘆くのならディストピアになるだろうが。
Rethinking Our Fear of AI: Is the Future Dystopian or Just Misunderstood?
Why are we so afraid of AI? Explore the cultural, political, and emotional roots of AI fear—and five real solutions to build a more human future
www.ainews.com
●Amazon/Kuiperが2度目の打ち上げ。今回は54衛星。トータルで3232の衛星打ち上げを計画。
●A2Aプロトコル(AIエージェントどうしを結ぶ通信規格)の開発と管理をLinux Foundationに移管。
Linux Foundation Launches the Agent2Agent Protocol Project to Enable Secure, Intelligent Communication Between AI Agents
Linux Foundation Launches the Agent2Agent Protocol Project
www.linuxfoundation.org
●T-Mobile、7月中にD2Cサービスをリリース。
T-Mobile to Launch T-Satellite Commercial Service Next Month
T-Mobile will be launching its nationwide satellite service to subscribers in late July, called T-Satellite.
www.satellitetoday.com
●LLMが人の感情を深く理解している可能性。EI(Emotional Intelligence)試験で高い得点。
Large language models are proficient in solving and creating emotional intelligence tests - Communications Psychology
Six Large Language Models outperformed humans on five ability emotional intelligence tests. ChatGPT-4 also successfully generated new test items for each test, ...
www.nature.com
⚫︎AI心臓病リスクをかなり前から評価できる。
AI Tool Detects Hidden Heart Risk in Routine Chest CT Scans
A new AI model finds calcium buildup in routine CT scans, helping flag heart disease early and potentially preventing cardiac events
www.ainews.com
コメント