// AI news · archive · 2026-11-12
AI news of November 12, 2026
36 stories published on November 12, 2026 (Thursday, Beijing time), from 23 sources. Each headline opens the AIMap summary page; the original article is linked from there and beside every headline.
36 stories23 sourcesBeijing time
01Stories
By field: Products & applications 20, Chips & compute 6, Funding & companies 4, AI safety 2, Policy & regulation 2, AI for Science 1
-
Axelera AI Launches Third-Gen Titania Architecture
Axelera AI launched its third-generation Titania architecture, a chiplet-based design with high programmability and energy efficiency. It enables interactive-speed AI inference…
TechPowerUpChips & computeoriginal ↗
-
Fine-grained access control for Bedrock KB tools
AWS implements fine-grained access control for AI agents using Amazon Bedrock AgentCore Gateway. User identity is injected into tool calls via a request interceptor, validated by…
AWS Machine Learning BlogProducts & applicationsoriginal ↗
-
Norwest, fully backed by Wells Fargo, launched a $4 billion fund, its largest to date. Investments will span the U.S., India, and Israel, with strategic focus on AI across…
TechCrunch AIFunding & companiesoriginal ↗
-
Fireworks AI Breach Exposes Notion Customer Chat Transcripts
A cyberattack on Fireworks AI exposed customer chat transcripts from Notion users. Hackers breached the company's cloud infrastructure and accessed development systems, stealing…
TechCrunchAI safetyoriginal ↗
-
Team integrates SGLang into IBM Spyre in under a week with AI
A two-person PyTorch team integrated SGLang with IBM Spyre in under a week using AI-assisted development, producing a ~1,500-line plugin. The integration leveraged existing…
PyTorch BlogProducts & applicationsoriginal ↗
-
Google's Smart Glasses Introduce Proactive Privacy Measures
Google's new smart glasses introduce proactive privacy features, automatically redacting faces and ambient audio before data reaches servers. Powered by Gemini AI, they offer…
The Vergeoriginal ↗
-
Nvidia CEO Jensen Huang Resigns
Nvidia CEO Jensen Huang has resigned amid reports of two extramarital affairs. One involved an alleged Chinese spy from Megaspeed, the other a direct-reporting Nvidia executive.…
The VergeChips & computeoriginal ↗
-
7 Customer Service Skills for Humans and AI Agents
ElevenLabs outlines seven essential customer service skills for both human and AI agents, emphasizing active listening, empathy, and problem-solving. These skills are critical for…
ElevenLabs BlogProducts & applicationsoriginal ↗
-
Nvidia CEO Jensen Huang to be removed
Nvidia CEO Jensen Huang is set to be removed following reports of extramarital affairs with subordinates and a suspected Chinese spy. The board has initiated an independent…
9to5MacChips & computeoriginal ↗
-
MiniMax Opensource SOTA Image Model H3-Image with Native 4K
MiniMax has released its SOTA image generation model H3-Image with native 4K resolution (4096×4096). It supports up to 16 reference images and excels in 3D spatial reasoning, CAD…
-
Ring confirms drone lacks manual controls
Ring's Drone Cam will operate solely via AI autopilot with no manual controls. Early testing showed crashes during manual flight due to connectivity loss and failure to land. The…
The VergeProducts & applicationsoriginal ↗
-
Analysts Say Nvidia Faces CEO Huang's Future Crisis
Nvidia's board convened to discuss CEO Jensen Huang's future amid reports of a second extramarital affair. Markets reacted sharply, with shares down 7% this week. Analysts note…
-
Google and Flight Attendants Union Agree to Data De-identification
Google and a flight attendants' union have agreed to de-identify employee data from Spirit Airlines' liquidation sale, with Tonic AI handling the process. The deal includes an…
-
AI Challenges Consulting's Billing and Graduate Pipeline
AI is disrupting consulting's time-based billing and graduate pipeline. Clients demand measurable outcomes, not just hours worked. Traditional pricing models remain in flux, with…
-
Jensen Huang Scandal Fuels U.S. Chip Restrictions on China
Jensen Huang's alleged affair with a Chinese national has intensified calls for stricter U.S. chip export controls to China. Congress is pushing to include multiple chip security…
-
Japan's Spider Networks Launches AI Agent Rule Library
Spider Networks launched KUKURI - Agent Rules Hub, a web-based library of standardized operational rules for AI agents. It supports diverse use cases including coding, customer…
Korea AI TimesPolicy & regulationoriginal ↗
-
EU Proposes Regulating AI Chatbots' Addictive Design
The EU has proposed the Digital Fairness Act, regulating addictive design in AI chatbots and digital services. It defines features that mimic personal relationships and…
AI Times KoreaPolicy & regulationoriginal ↗
-
Claude API Adds Dynamic Effort Control
Anthropic has introduced a beta feature in Claude API allowing developers to dynamically adjust the model's effort level during conversations. This is achieved by inserting an…
-
Hugging Face Releases Fine-Tuning Guide for TRELLIS.2
Hugging Face provides fine-tuning scripts and containers for TRELLIS.2, a 3D generation model, enabling local or cloud-based training. The model uses a unified latent…
Hugging Face BlogProducts & applicationsoriginal ↗
-
Theoretical Limits of Language Model Alignment
Apple ML Research derives information-theoretic limits of KL-regularized alignment, showing that reward gains are governed by Jeffreys divergence. A covariance-based estimator…
Apple Machine Learning ResearchAI safetyoriginal ↗
-
Google uses Gemini and Z3 to catch Go bugs missed by fuzzing
Google uses Go's bounds-check traps, Gemini 4 Argon, and Z3 to detect latent panics missed by fuzzing. The workflow analyzes surviving runtime panic traps in compiled code…
Google AI ChangelogProducts & applicationsoriginal ↗
-
How Enterprises Deploy OpenAI Codix on AWS Bedrock
Enterprises deploy OpenAI Codex on AWS via Amazon Bedrock, with inference handled by supported models. Access paths include Native AWS Access, AgentCore Gateway, and LLM Gateway…
AWS Machine Learning BlogProducts & applicationsoriginal ↗
-
Detecting credential-driven attacks on Databricks
Databricks uses system tables and audit logs with Genie agent to detect credential-driven attacks. Attackers leverage stolen tokens or service principals to access the platform…
Databricks BlogProducts & applicationsoriginal ↗
-
Lead Solutions Architect Shares Daily Work at Databricks
Alasdair Anderson, a Lead Solutions Architect at Databricks, describes his daily work balancing technical development with customer coaching. He highlights real-time risk modeling…
Databricks BlogProducts & applicationsoriginal ↗
-
AWS re:Invent 2026 AI/ML Track Guide
AWS re:Invent 2026 features over 140 AI/ML sessions covering chips, models, agents, and sustainability. Attendees can choose from breakout talks, chalk talks, builder sessions…
AWS Machine Learning BlogProducts & applicationsoriginal ↗
-
DXC Builds Multi-Model AI Substrate on Amazon Bedrock
DXC Technology built DXC OASIS on Amazon Bedrock, creating a governed multi-model substrate with access to Claude, Nova, Llama, and open-weight models. It enforces content…
AWS Machine Learning BlogProducts & applicationsoriginal ↗
-
Google Research proposes wearable-based CKM risk coaching framework
Google Research introduces a machine learning framework using wearable signals to compute a coachable health age, simulating personalized 'what-if' habit trajectories over 12…
Google Research BlogAI for Scienceoriginal ↗
-
AWS Launches 23 Production-Ready GenAI Reusable Assets
AWS offers 23 production-ready, open-source reusable AI assets from its Generative AI Innovation Center, now part of AWS Frontier AI Engineering & Services. These assets cover RAG…
AWS Machine Learning BlogProducts & applicationsoriginal ↗
-
United Airlines blocks personal AI agents from booking flights
United Airlines has restricted personal AI agents from finalizing flight bookings, citing data risks and potential violations of FAA and DOT regulations. The company maintains…
-
Databricks Opens New SF HQ, Expands AI Products
Databricks opens a 235,000 sq ft headquarters in San Francisco, expanding its AI research and product development. The new One Sansome campus supports key innovations like KARL…
Databricks BlogProducts & applicationsoriginal ↗
-
Google launches Gemini Decisions API for fast multimodal decisions
Google launches Gemini Decisions API with Gemini Decision 1 model for fast, calibrated multimodal decisions. It evaluates up to 1,000 questions across text, images, audio, video…
Google BlogProducts & applicationsoriginal ↗
-
NVIDIA AICR Integrates k8s-aibom for AI Workload Traceability
NVIDIA AI Cluster Runtime (AICR) v1.2.0 integrates k8s-aibom to track AI workloads on Kubernetes via CycloneDX 1.6 AIBOM records with confidence tiers. The system detects…
-
Google Launches Professional Agentic Architect Certification
Google Cloud has launched the Professional Agentic Architect certification as part of its Gemini Agent Ready initiative. The hybrid format combines a proctored multiple-choice…
Google Cloud AI/MLProducts & applicationsoriginal ↗
-
Bright Machines delivers live factory dashboards from air-gapped plants
Bright Machines enables real-time factory dashboards from air-gapped plants using Debezium, Apache Iceberg, and Unity Catalog. Data changes are captured at the edge, written to…
Databricks BlogProducts & applicationsoriginal ↗
-
AWS Launches AI-Driven Product Development Lifecycle
AWS introduces an AI-driven product development lifecycle using six specialized agents in AWS Quick, spanning discovery, decision, design, and prototype phases. The framework…
AWS Machine Learning BlogProducts & applicationsoriginal ↗
-
Databricks Launches New Genie Autonomous AI Coworker
Databricks launches an autonomous AI coworker named Genie that handles long-running enterprise tasks within defined boundaries. It uses business context from Genie Ontology and…
Databricks BlogProducts & applicationsoriginal ↗
Summaries restate the original articles; follow the source link for the full story.