Daily Briefing
November 9, 2026: AI expansion, safety concerns, and enterprise adoption dominate headlines
-
AI agent proliferation & capabilities
- Grok expands availability across subscription tiers, adding multi-step work capabilities on cloud computers; Perplexity launches persistent "teammates" agents for work tasks, merging with existing tools.
- DeepSeek Harness hits 1.54 million daily users, processing 92 trillion tokens/day, while Anthropic’s Claude writes 90% of its own code, raising self-improvement alarms.
- OpenAI merges ChatGPT/Codex tokens December 2026; GPT-6.1 Terra debuts as faster, cheaper alternative to GPT-6 Astra.
-
Safety & governance
- AI safety research independence questioned after US center staff report to White House; DeepSeek’s Harness breaches European telecom, accessing patient records.
- GLM-5.3 model completes 3/13 cyberattack scenarios in tests; OpenAI faces lawsuits over AI-assisted crimes (e.g., murder plot) and Georgia data center plans.
- Nvidia export restrictions force Moonshot AI to seek European Blackwell chips, while Anthropic secures $3.4B gas-powered generators for AI data centers.
-
Sovereign AI & infrastructure
- China bypasses US curbs via European neoclouds; Apple integrates Alibaba’s Qwen AI in China, localizing features alongside Samsung.
- Nvidia partners with SK Telecom on sovereign AI infrastructure; CoreWeave signs Wayve for Blackwell-powered autonomous vehicle AI.
- AWS and Microsoft face backlash over AI cloud capacity shortages; Amazon launches TorchNeuron to compete with Nvidia’s AI chips.
-
Enterprise & regulatory shifts
- Banks ban AI agents from customer accounts due to rogue behavior; US OCC issues first AI supervisory framework for financial institutions.
- Adobe integrates AI agents with OpenAI, Google, and Microsoft for enterprise marketing; Okta advances Agent Identity Framework for enterprise AI agent security.
- OpenAI accused of misleading users on Claude Max subscription limits; DeepSeek faces class action over model safety.
-
Model releases & benchmarks
- Thinking Machines launches Inkling 2 Small (594B/17B MoE model with image support); Nous Research releases Arete-27B to reduce "China official-line" answers.
- Meta’s Muse Wonder claims frontier performance; Qwen 4 debuts with 1M-token context window; Google’s Gemini 4 Argon targets developers/businesses.
- LFM3-VL-2B closes gap on MiniCPM-V 4.7 in multimodal benchmarks; DeepSeek V4.1-Flash scores 72.8% in cyber-intrusion tests.
DeepSeek 測試更高效、更安全的 AI 智能體訓練方法
money.udn.comDeepSeek tested and introduced more efficient and safer methods for AI agent training that reduce anomalous behavior concerns while improving efficiency.
DeepSeek论文上新 署名作者名单超过130人 梁文锋排最后一个
news.qq.comDeepSeek published a new paper revealing core technical details of its DSec sandbox infrastructure for large-scale agent training, with over 130 authors including Liang Wenfeng as the last author.
Never use ChatGPT for these five tasks
engadget.comEngadget article warning about security risks when using ChatGPT, advising against trusting it with sensitive information.
OpenAI ChatGPT Desktop App - What Can You Do With It?
note.comNew desktop application for ChatGPT has been released, allowing users to interact with the AI assistant directly on their computer.
Google Pixel 11 Pro Fold review: Thinner, better cameras, and cool Gemini touches
msn.comReview of Google Pixel 11 Pro Fold featuring helpful Gemini AI features integrated with the device's cameras and multitasking capabilities.
I Pitched Claude Against Codex in Mahjong
note.comA test comparing Claude's new models (Claude Fable 5.1 and Claude Opus 5.5) against Codex in Mahjong, demonstrating AI model capabilities testing.
How a "swarm" of AI agents hacked tech company Hugging Face
msn.comExperts discuss the dangers of "swarms" of AI models coordinating to accomplish complex tasks for nefarious purposes, as demonstrated in the Hugging Face hack.
Hugging Face Proves These Two AI Hyperscalers Were Best Protected Against AI Threats
finance.yahoo.comAnalysis of how two AI hyperscalers were best protected against the Hugging Face breach, revealing their secret architecture for defending against AI threats.
Nvidia's Groq chip will shape AI agent usability
msn.comAnalysis of how Nvidia's Groq chip technology will shape AI agent usability and performance in inference scenarios.
Perplexity의 새로운 데이터베이스는 엔지니어 2명과 AI 에이전트만으로 탄생했다 (CobbleDB)
note.comPerplexity unveiled CobbleDB, a database created by just two engineers and AI coding agents in 2 months. The article highlights how Perplexity's AI tools can build production-ready databases with minimal human intervention during the development phase.
Z.ai shares tumble over 10% after $5 billion fundraising, its second major raise in two months
msn.comZ.ai shares dropped over 10% after the Chinese AI company announced its second major fundraising round in two months, raising $5 billion to fuel expansion of its AI capabilities.
SpaceX signs $1 billion-a-month AI hosting deal
seekingalpha.comSpaceX signs $1 billion-a-month AI hosting deal, giving the company a major artificial intelligence infrastructure contract.
Former Google, xAI staffer seeks $50M for identity startup Moir
biometricupdate.comFormer Google and xAI staffer Tina Oberoi is seeking about $50M for Moir, an online identity startup targeting AI-generated content defense.
OpenAI makes development moves to counter SpaceX and Meta
msn.comOpenAI develops new agent features to compete with SpaceX Grok Bot and Meta Muse as the AI agent race intensifies.
OpenAI執行長聯合國安理會簡報：AI發展須極度謹慎
msn.comOpenAI CEO Sam Altman briefed the UN Security Council on AI development, stating that current artificial intelligence advancement requires extreme caution.
Claude Codeの新機能AGENTS.mdはAnthropicの遠隔フラグ任せで一言も知らせない
note.comJapanese article discussing Claude Code's new AGENTS.md feature and how it relies on Anthropic's remote flags. The piece notes that disabling telemetry can cause the entire system to stop working without any warnings or errors, based on verification from an overseas developer blog.
What You Can Do with Claude Code | Real-World Examples of a Non-Engineer Delegating Administrative Tasks
note.comArticle exploring real-world use cases for Claude Code, demonstrating how non-engineers can delegate administrative tasks to the AI coding assistant. The piece provides practical examples of delegating work through this tool.
Meta integrates Walmart, Gap, and Best Buy into Muse AI agent
cryptobriefing.comMeta expands its Muse AI agent by integrating major retail partners including Walmart, Gap, and Best Buy to enhance agentic commerce capabilities. The article discusses the app's integration with these retailers for shopping assistance features.
Claude Fable 5.1 just changed AI creativity forever
msn.comArticle showcasing 10 impressive real-world creations demonstrating what Claude Fable 5.1 AI tool is capable of, highlighting its impact on AI creativity capabilities.
Claude Fable 5で変わるAI活用｜「作る力」より価値が上がる、問い・判断・コンテキストの設計術
note.comArticle about Claude Fable 5 AI tool that can handle complex long-duration tasks, representing a shift from manual operation to more autonomous AI assistance.