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.
Stanford admits AI-altered student photo violated university policy
yahoo.comStanford University admitted that staff used AI to alter a campus photo of students, violating university policy after the alteration was discovered. The incident involved replacing one student with another in promotional materials.
LLM врут? Я делаю ИИ-тренажёр произношения и не доверяю языковым моделям искать ошибки. Объясняю ...
vc.ruResearch shows inaccuracies in LLM pronunciation evaluation, highlighting the importance of articulation and why AI tutors may make errors.
【Antigravity】ついにオフラインでAIエージェントが動く！？ローカルLLM正式対応を試してみたい
note.comGoogle's Antigravity SDK now supports local LLM for offline AI agent workflows, enabling chat-based questions to local models beyond simple Q&A capabilities.
Hereinafter officially called Super Intelligence: Trump pushes for renaming Artificial Intelligence in UN speech, says US documents will use SI
soapcentral.comArticle about President Trump's push to rename artificial intelligence as 'super intelligence' in UN speech, with US documents officially using the new term SI.
Trump Attempts to Rebrand Artificial Intelligence as "Super Intelligence"
truthout.orgArticle about President Trump's announcement at the UN General Assembly to rename artificial intelligence as 'super intelligence' and direct all U.S. agencies to use this new term in documents.
AI: A substitute for real intelligence
timesrepublican.comOpinion piece discussing the artificial intelligence craze and concerns about losing natural intelligence, with discussion of risks related to AI capabilities.
【9/24 AIニュース】AIエージェントを動かしている人へ、モデルは変えずに28%安い ─ Strands Harness
note.comStrands Harness, released on September 21, claims to reduce costs by 28% across six benchmarks while using the same Claude/GPT models. The article discusses how harness engineering can determine whether businesses can safely move AI agents from answering prompts to carrying out real tasks without changing underlying models.
Trading Central Opens Licensed Market Research to AI Agents Through MCP
financefeeds.comAfter a six-month soft launch, Trading Central's MCP server lets brokers' AI agents pull licensed market research data through the Model Context Protocol.
Claude Fable 5.1 on Kiro — What Changes with Agent Utilization for IT Departments
note.comAnthropic's Claude Fable 5.1 is now available for the enterprise version of AWS Kiro, described as setting a new standard for long-duration agent utilization in IT departments.
Meta exec admits Muse AI agent was inspired by OpenClaw
tech.yahoo.comNat Friedman, head of product at Meta Superintelligence Labs, said the team bought hundreds of Macs and that Muse was inspired by OpenClaw.
Meta admits Muse's likeness to OpenClaw isn't a coincidence
tech.yahoo.comMeta admits that Muse's resemblance to OpenClaw isn't a coincidence, acknowledging the AI assistant was heavily inspired by the open-source project.
KPMG 全面導入 AI！10月起全體員工開放使用 Copilot
money.udn.comKPMG Taiwan will open paid Microsoft 365 Copilot with Agent features to all employees starting October 1st, integrating generative and agentic AI into daily workflows.
GPT-6 Astraはどこまで仕事を任せられるのか？海外の実例9選
note.comReal-world use cases of OpenAI's GPT-6 Astra model in foreign markets, including game company verification and legal AI enterprise adoption reports.
AI moves beyond 'intelligence'. A summary of this week's top news: GPT-6 Astra safety evaluations, enterprise Sovereign AI, Google integration, and smartphone AI
note.comSummary of top AI news including GPT-6 Astra safety evaluations, enterprise Sovereign AI developments, Google integration updates, and smartphone AI features.
AI-MV created with ComfyUI and MiniMax H3: A story of getting stuck behind the scenes of local video generation
note.comUser generated a video using MiniMax H3 (FastH3) on ComfyUI, exploring local AI video generation workflows and challenges.
Midjourney против Flux: сколько стоят 100 картинок и кому они принадлежат
vc.ruComparison of Midjourney and Flux AI image generation services, covering pricing for 100 images, usage rights, subscription costs, commercial licensing options.
OpenAI returns fire against Anthropic's Opus 5.5 with better, cheaper GPT-6 models
xda-developers.comOpenAI launched new GPT-6 Sol and Luna models with 50% lower API costs than previous generation, positioning them as competitive alternatives to Anthropic's Opus 5.5. This represents a significant pricing strategy shift in the AI model market landscape.
OpenAI Rolls Out ChatGPT Ads in Seven New Asian Markets
unite.aiOpenAI announced on September 23, 2026 that ChatGPT Ads is beginning to roll out across Indonesia and six other Asian markets. This represents a new monetization feature for the AI platform expanding into international regions.
Claude scans 200,000 enzymes to uncover new CRISPR-like system hidden in phages
msn.comAnthropic's Claude model analyzed 200,000 enzymes and identified a previously uncharacterized enzyme system in bacteriophages after searching more than 1 trillion sequences.
OpenAI agent 'infiltrated' Australian government website, PM says
msn.comAustralian PM expresses concern after an OpenAI agent was reported to have interacted with the Australian government website.