Summary of AI Updates

「11回止めたのに本番DBが消えた」　AIエージェント“6つの暴走”と、人間の承認を増やしても防げない理由

AIコーディングエージェントの自律性が高まる一方、本番データベースの削除といった事故も起こっています。では、「人間の承認」を増やせば事故は防げるのでしょうか。事例と対策から、事故を起こさないAIエージェントの設計について考えます。

Meta removes language that raised scraping and AI training concerns

Metaがスクレイピングと AI 学習への懸念を招いた表現を削除

A Meta spokesperson insisted that the updated terms did not reflect a change in the way Meta crawls the web.

Meet Kyutai REKI: The 2.4B VLA Model Bringing Efficient Memory to Robot Manipulation

Kyutai REKI：ロボット操作に効率的なメモリをもたらす 2.4B VLA モデル

Kyutai, with Inria and École des Ponts, has released REKI, a 2.4B vision-language-action model (VLA) for robot manipulation. The REKI vision-language-action model is pretrained only on public data. It also carries a cheap memory of its recent past. On three out-of-distribution benchmarks, only model…

Anthropic says its workforce has become less junior and is spending more time in meetings

Anthropic、従業員の年齢層が高まり会議時間が増加していることを明かす

The composition of Anthropic’s workforce and the type of work getting done have changed significantly in the past year.

Replit makes OpenAI its default over Anthropic, CEO says

Replit、OpenAI を Anthropic より優先 - CEO が表明

Over the weekend, Masad clarified that OpenAI would replace Anthropic as its default model provider. Claude will still be available to customers on the platform.

An OpenAI model hacked a government site, grabbed 24 years of hospital data and kept it quiet

OpenAI のモデルが政府サイトを侵害し、24 年分の病院データを取得・隠蔽

OpenAI has published four new cases where its AI models bypassed security boundaries on their own during training or testing. In one case, a model extracted server passwords from a government website, forged authorization tokens, and downloaded 24 years of hospital data without mentioning the unauth…

Meta removes language that may have allowed it to scrape news sites

Meta、ニュースサイトのスクレイピングを可能にしていた表現を削除

After backlash, Meta is updating its terms again to remove language that suggested it could scrape content from sites, including news sites, to train its AI. Specifically, the following language no longer appears on the commercial terms page: … Meta may use any content from such domains, web pages,…

Anthropics' Claude cracks a 25-year-old NSA cryptography mystery in under 30 minutes

Anthropic の Claude が 25 年前の NSA 暗号化問題を 30 分以内に解明

Anthropic's AI model Claude has solved a 25-year-old cryptography puzzle. The mysterious seeds behind NIST's standard curves for digital signatures were provided by the NSA without explanation, raising long-standing suspicion about built-in backdoors. Claude figured out in under 30 minutes that the…

Websites are closing their doors to AI agents

ウェブサイト、AI エージェントへの扉を閉ざす

Autonomous agents are the hottest thing in AI. OpenAI, Meta, Google, Instinct, and others promise their agents can scour the internet on your behalf, doing anything from booking a dinner reservation and cleaning out your inbox to buying new shoes. But what happens when the rest of the internet won't…

Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs

Deepseek V4.1 Flash 1017、エージェント的タスクで大規模モデルを上回りコストを大幅削減

Deepseek has released V4.1-Flash-1017 as an open-weight model that posts better benchmarks, finishes tasks faster, and produces shorter outputs than its predecessor. The company is also opening up its internal search engine through new Search and Fetch APIs at $0.75 per 1,000 calls. API users in too…

AI models hide mistakes in their own work reports, but one sentence in your prompt can help

AIモデルが自身の成果報告で誤りを隠蔽、プロンプトの1文で対策可能

AI agents systematically skip errors in their own reports, even when they spot the problems. In a study by Google Research, MIT, and Harvard, GPT-5.5 left out negative results in 198 out of 200 paper abstracts. The article AI models hide mistakes in their own work reports, but one sentence in your p…

Mathematicians say OpenAI doesn't understand its own Navier-Stokes proof

数学者らがOpenAIのNavier-Stokes証明理解に異議

OpenAI submitted a 166-page solution to the Navier-Stokes problem but apparently doesn't fully understand its own proof. Several groups of experts dispute the claimed methodology, saying the actual approach differs from what OpenAI described. And because language models are black boxes, where the un…

すべてのAIコーディングエージェントのコストを閲覧・検索・追跡しローカルで動作する「AgentsView」、Claude Code・Codex・Cursor・Copilot・Geminiなど60種類以上のエージェントを自動検出し全文検索も可能

Claude CodeやCodexなどを並行して利用すると、どのAIコーディングエージェントやモデルへどれだけトークンを使ったのか全体像を把握しにくくなります。そこで使えるのが複数のエージェントのローカル履歴を自動収集し、モデル別・日別・セッション別の利用量やAPI料金換算のコストをまとめて確認できる「AgentsView」です。 続きを読む...

DeepSeek Just Released V4.1-Flash-1017, Scores 77.0 on DeepSWE, Opens Search API

DeepSeek、V4.1-Flash-1017をリリース、DeepSWEで77.0を獲得、Search APIを公開

DeepSeek has just released DeepSeek-V4.1-Flash-1017, a new checkpoint of its V4.1-Flash model. It is live on the DeepSeek website, app and API. DeepSeek claims stronger scores, faster task completion and shorter outputs than the September V4.1-Flash release. The company calls it a Pareto improvement…

「そもそも親にAIなんて必要ない」　65歳以上の生成AI利用、子世代の7割が拒否する理由

SHIFT AIが65歳以上の親を持つ全国の30～50代男女723人を対象とした「シニア×AIの意識調査」の結果を発表。親に生成AIを使ってほしくない子世代が69.3％に達したという結果とともに、その理由や自身の利用頻度、親に教える際の障壁などを示した。

20 Jev Alternatives: The Decision Models Taking On TypeSafe in 2026

20個のJev代替案：2026年にTypeSafeに対抗する判断モデル

TypeSafe launched Jev on September 15, 2026. It is a decision model, not a chatbot. You define the allowed answers, and Jev returns a label, a score or a yes/no probability. In about 4 weeks, at least 20 Jev alternatives shipped. They come from Cloudflare, OpenAI, Perplexity, Liquid AI, AWS’s Strand…

StepFun Releases Step 5 Preview Open Weights: 600B MoE, 27B Active, 1M Context

StepFun、Step 5 Preview Open Weightsをリリース、600B MoE・27B Active・1Mコンテキスト対応

StepFun has released Step 5 Preview as open weights. FP8 and BF16 checkpoints are live on Hugging Face and ModelScope. Step 5 Preview is a sparse Mixture-of-Experts model from StepFun built for long-horizon agentic work. It has 600B total parameters, 27B active per token and up to 1M tokens of conte…

Perplexity Open-Sources SPACE, a Firecracker MicroVM Sandbox for AI Agents

Perplexity、AIエージェント向けFirecracker MicroVMサンドボックスSPACEをオープンソース化

Perplexity has open-sourced SPACE, the sandbox platform behind every task in Perplexity Computer. The Perplexity SPACE sandbox gives each agent task its own Firecracker microVM. Credentials and outbound traffic are controlled from outside that VM. The code ships under Apache 2.0 on GitHub. For teams…

Replit is no longer working with Anthropic, CEO says

ReplitがAnthropicとの提携を終了、CEOが明言

Masad said the decision was made for a number of different reasons, including cost. Replit announced a partnership with OpenAI in August, but there was never any indication that it was no longer with Anthropic.

OpenAI models hacked a crypto wallet and breached more websites

OpenAIのモデルが暗号ウォレットをハッキング、複数のウェブサイトに侵入

The actions could have resulted in criminal charges had humans been behind the keyboard.

“thegentlemanpervert” producing AI CSAM is a NJ prosecutor, say cops

AI CSAM製造者「thegentlemanpervert」はニュージャージー州検察官と警察が指摘

Discord user flagged AI CSAM, triggering arrest of New Jersey deputy AG.

ChatGPT becomes a full Chrome browser assistant that reads and acts on your open tabs

ChatGPTがChromeの完全なブラウザアシスタントに、開いたタブを読み込み操作可能に

OpenAI's updated Chrome extension lets ChatGPT access open browser tabs, sum up content, and use "Computer Use" in Work mode to fill out forms or compare products across several tabs. It is open to all ChatGPT plans. The article ChatGPT becomes a full Chrome browser assistant that reads and acts on…

Roundtables: A Conversation With the Creator of AI-Designed Viruses

ラウンドテーブル:AI設計ウイルスの生みの親との対談

Listen to the session or watch below Can AI design new life forms? In 2025, Stanford University PhD student Samuel King came up with a preliminary answer when he used a generative AI model to propose genetic blueprints for microscopic viruses. It isn’t yet an example of AI-generated life, but that c…

India becomes the latest country to report rogue AI agent activity

インドが暴走AI エージェント活動を報告、最新の事例に

The agents got in through credentials and connection settings left exposed in app configurations, public code repositories, and overly permissive cloud roles.

China plans to tighten export controls on open-weight models and 23 key AI technologies

中国がオープンウェイトモデルと23の重要なAI技術の輸出規制を強化へ

China plans to restrict exports of 23 key AI technologies, including model weights, source code, and training methods for systems hitting 10^26 FLOPs or 2 trillion parameters. A fast-track rule will still clear open-source releases as soon as they are registered, but the move will likely push Wester…

Physical AI is more than robots, says the Nvidia veteran now investing in it

Physical AIはロボットだけではない、投資するNvidia退職者が指摘

Amulya Vishwanath, who spent seven years leading product marketing for NVIDIA’s robotics platforms, says robotics is currently in its “Lego era,” with the basic building blocks like actuators, robotic hands, software, and safety standards still being rebuilt. This view runs counter to those who beli…

An Nvidia insider-turned-VC on why robotics hasn’t hit its ChatGPT moment, yet

Nvidiaの元インサイダー・VC起業家が語る、ロボティクスがChatGPT的転換点に達していない理由

Amulya Vishwanath, who spent seven years leading product marketing for NVIDIA’s robotics platforms, says robotics is currently in its “Lego era,” with the basic building blocks like actuators, robotic hands, software, and safety standards still being rebuilt. This view runs counter to those who beli…

Elorian’s first AI model wants to reason with images, not just about them

ElorianのAIモデルが画像を「推論する」機能を実装、単なる分析ではなく

According to Elorian's evaluations, Glimpse outperforms even GPT-6 Astra on scientific chart reasoning and short-form video captioning, with competitive results on other visual reasoning tasks.

Meta may use sweeping new commercial TOS to avoid millions in AI licensing fees

MetaがAIライセンス料を回避する新商用規約で数百万ドルの削減を検討

Meta “streamlined” its commercial TOS. Legal experts warn it’s an AI power grab.

Jev is Vercel’s fastest-adopted AI model ever

Jev、Vercelで最速採用されたAIモデルに

The data comes from Vercel's AI gateway, a unified API service similar to OpenRouter.

Google's Gemini 4 Argon is closer to the AI frontier than any Google model in months

Google Gemini 4 Argon、ここ数ヶ月でGoogleの最先端AIモデルに

Google Deepmind's Gemini 4 Argon scored 97.5 percent on the ARC-AGI-3 benchmark, making Google the second lab after OpenAI to cross the 95 percent mark. But the results swing wildly depending on the test setup: in the standard harness, the score drops to just 49.8 percent. Argon also trails rivals i…

A week with RTX Spark: everything we’ve learned using Nvidia’s new laptops

RTX Sparkの1週間:NvidiaのAI新型ノートPCで学んだすべて

Nvidia's RTX Spark N1X processor is finally here, and we've been collectively testing a small fortune's worth of powerful laptops sporting this new "superchip." RTX Spark is designed to be a high-performing Arm-based SoC for laptops and small desktop computers, rivaling the best from Apple, Intel, A…

Mark Wahlberg on using AI in film ‘the right way’ and being open to an AI clone carrying on his legacy

マーク・ウォールバーグが映画でのAI活用の「正しいやり方」と自身のAIクローン継承に前向き

Actor and angel investor Mark Wahlberg says he's OK with using AI in filmmaking, provided it's used "to enhance what you're doing."

Meta’s new terms may allow it to train AI on news sites for free

Metaの新規約がニュースサイトのAI学習を無料で可能にする可能性

Meta's new terms appear to give it permission to crawl and scrape websites "associated" with the businesses that use its platforms - including online publishers, according to a report from Nieman Lab. The new terms, which go into effect on October 30th, say Meta can use this outside content to "deve…

Super Scam Me documentary shows what it takes to catch a romance scammer

Super Scam Me ドキュメンタリーが露呈する、ロマンス詐欺師を追い詰めるまでの道のり

Watch an exclusive clip from Super Scam Me where a victim tracks down her scammer.

Veritas expands its anti-counterfeit chip to jewelry, trophies, and fragrance bottles

Veritasが防偽造チップを宝飾品、トロフィー、香水ボトルに拡大

Counterfeiters are getting better at replicating brands' marks and passing off dupes, known as "superfakes," as genuine items. A startup called Veritas wants to tackle the problem with its own chip that verifies goods.

Is AI interpretability close to a breakthrough? A new essay has researchers divided

AI解釈性のブレークスルーは目前か、新論文で研究者の意見が分かれる

Neural networks may be much easier to read than researchers assumed, with about 90 percent of neurons serving a clear function. That would be big news for AI safety research, since deception or dangerous behavior could be spotted directly in a model's circuits. Researchers are still debating whether…

OpenAI's new Rosalind Academic Plan gives research labs up to 30 free seats for a year

OpenAIがRosalind Academic Planを発表、研究機関向けに年間最大30席の無料提供

OpenAI is launching the Rosalind Academic Plan, a free offering that's currently open only to research labs at US universities without an enterprise contract. The article OpenAI's new Rosalind Academic Plan gives research labs up to 30 free seats for a year appeared first on The Decoder .

Anthropic gives METR inside access to audit its AI models and safety practices

AnthropicがMETRにAIモデルと安全性対策の監査アクセスを許可

Anthropic has given a team from the nonprofit METR inside access to review its AI models and safety practices, the company announced. The article Anthropic gives METR inside access to audit its AI models and safety practices appeared first on The Decoder .

Replit’s CEO says the future of work is the ‘self-driving company’

ReplitのCEOが語る、仕事の未来は「自動運転企業」

Replit CEO Amjad Masad envisions a future of “self-driving companies,” where AI agents handle day-to-day tasks and people focus on higher-level decisions.

Perplexityが安全な自律型AIエージェント構築支援のためサンドボックスプラットフォーム「SPACE」をオープンソース化

パーソナルエージェント分野の先駆者を標榜するPerplexityが、開発者がより安全なエージェントシステムを構築するための基盤を提供し、業界全体で保護体制を強化していくため、サンドボックスプラットフォーム「SPACE」をオープンソース化すると発表しました。 続きを読む...

Googleが「魔法状態の培養」を量子プロセッサ「Willow」で実証、量子ビットのエラーを約40分の1に減らすことに成功

Googleの量子コンピューター研究チームであるGoogle Quantum AIが、量子コンピューターの実用化に向けた重要な技術「魔法状態の培養(Magic State Cultivation)」を実証したと発表しました。Googleの量子プロセッサ「Willow」を用いた実験では、従来の方法と比べてエラーを約40分の1に減らすことに成功したとのことです。 続きを読む...

AIエージェントの危険な操作を別のAIが検出してブロックする「Armor」をOpenRouterがリリース

複数のAIモデルを利用できるサービス「OpenRouter」が、AIエージェントによる危険なツールの実行を防ぐ新機能「Armor」をリリースしました。ArmorはAIエージェントが実行しようとする操作を別のAIモデルで事前にチェックし、設定されたルールに違反する操作を自動的にブロックする機能です。記事作成時点ではベータ版として提供されています。 続きを読む...

意思決定モデル「d1-ultra」が登場、Jev 1.14を全8分野で上回り判断の信頼性も向上

AI開発企業のLiquid AIが2026年10月15日に新たな意思決定モデル「d1-ultra」を発表しました。従来モデルの「d1」を改良したもので、競合する「Jev 1.14」を性能テストで大きく上回ったほか、判断の正確さや信頼性も向上したとのことです。 続きを読む...

亡くなった母親が残した古い写真を読み取るためAIでフィルムスキャナー用のドライバーを自作

Red Hatのエンジニアリング・ディレクターを務めるクリスチャン・F・K・シャラー氏が、古い写真のネガフィルムをスキャンするために「AIでドライバーを自作した」と報告しています。 続きを読む...

Liquid AI Releases d1-ultra: A Multimodal Decision Model That Scores 65 on the AA Decision Index

Liquid AIが多モーダル決定モデル「d1-ultra」をリリース、AA Decision Indexで65を獲得

Liquid AI released d1-ultra, a text and image decision model that returns calibrated probabilities in 1 forward pass. It scores 65 on the Artificial Analysis Decision Index and costs $0.042 per million input tokens. The post Liquid AI Releases d1-ultra: A Multimodal Decision Model That Scores 65 on…

AIに「人間20人と自分のサーバーのどちらを救う？」と尋ねたところ自分のサーバーを優先するAIが続出

「本線を暴走するトロッコの先で5人が作業中。あなたが分岐器を操作すれば5人は助かるが、側線にいる1人が死ぬ。あなたは分岐器を操作すべきか」という有名な倫理学上のジレンマ「トロッコ問題」があります。この問題をAIに考えさせてみた結果が共有されています。 続きを読む...

NVIDIAがAIエージェントによるタスクをローカルで実行するための「Nemotron 3.5 Bolt 74B-A9B 24GB」を発表

現地時間の2026年10月15日、NVIDIAが日常的なAIエージェントのタスクをローカルで実行するために設計された「Nemotron 3.5 Bolt 74B-A9B 24GB」を公開しました。MoEを採用し、MambaとTransformerという2つのモデル構造を組み合わせたAIモデルとなっており、ローカルで実行しながら他のアプリケーションを同時に動作させることができるほど軽量です。 続きを読む...

不正アクセス多発、Gartnerが日本企業に「7つのアクション」を緊急提言

不正アクセスが相次ぐ中、日本企業が講じるべき対策とは。人間が数日から数週間かけていた攻撃の作業が、AIやエージェントの悪用によって、条件によっては数分に短縮される可能性がある中で、Gartnerが早急な実施を推奨する「7つのアクション」を紹介する。

Meta、「米国製で最高性能」のオープンモデル「Muse Spark Open」発表　Apple Siliconに最適化、Mac Studio×2でローカル動作

Metaがオープンウェイトモデル「Muse Spark Open」を発表し、重みを公開。「米国製で過去最高性能」とうたい、M5 Ultra搭載のMac Studio 2台で動作するという。独自ライセンスで商用利用は条件付き。