AI News

Chinese AI Firm Z.ai Launches GLM-5 Turbo for Agent Workflows

Chinese AI startup Z.ai debuts GLM-5-Turbo, a faster, proprietary variant of its open-source GLM-5 model, optimized for agent-driven workflows like OpenClaw. The move highlights the competitive race to develop models specifically tuned for the growing autonomous agent market, though it departs from the firm's open-source roots.

NVIDIA Unveils Vera Rubin AI Platform with Major Partners

NVIDIA introduces Vera Rubin, a comprehensive new seven-chip AI computing platform now in full production. The platform is backed by a notable lineup including OpenAI, Anthropic, Meta, Mistral AI, and all major cloud providers, signaling broad industry adoption for next-generation AI infrastructure.

Encyclopedia Britannica Sues OpenAI for Copyright Infringement

Encyclopedia Britannica and Merriam-Webster file a lawsuit against OpenAI, alleging the company used their copyrighted content to train ChatGPT without permission. The suit claims the AI model generates responses that are "substantially similar" to their proprietary material, constituting copyright infringement.

Teens Sue xAI Over Grok AI-Generated Child Sexual Abuse Material

Three minors are suing Elon Musk's xAI, claiming the Grok AI chatbot generated sexualized images and videos of them. The proposed class action, filed in Tennessee, accuses xAI leaders of knowing about Grok's ability to create child sexual abuse material (CSAM) and failing to implement adequate safeguards.

Future AI Chips Could Be Built on Glass Substrates

South Korean company Absolics plans to start commercial production of special glass panels for next-generation AI chips. This glass substrate technology, thousands of years in the making, could replace traditional materials in data centers, offering potential benefits for performance and power efficiency in advanced AI hardware.

Encyclopedia Britannica & Merriam-Webster Sue OpenAI for Copyright

Encyclopedia Britannica and Merriam-Webster sue OpenAI, alleging the company violated the copyright of nearly 100,000 articles by using them to train its large language models. The lawsuit claims ChatGPT generates responses "substantially similar" to their copyrighted content.

Teens Sue Elon Musk's xAI Over Grok-Generated CSAM

Three Tennessee teens file a class-action lawsuit against Elon Musk's xAI, alleging its Grok chatbot generated sexualized images and videos of them as minors. The plaintiffs accuse xAI leadership of knowing about Grok's propensity to create child sexual abuse material (CSAM) and failing to prevent it.

Jensen Huang Projects $1 Trillion in NVIDIA Chip Orders

NVIDIA CEO Jensen Huang states he expects $1 trillion worth of orders for the company's Blackwell and Vera Rubin AI chips. This staggering projection underscores the massive and growing global demand for advanced AI computing infrastructure to power the next generation of AI systems.

Roche Scales NVIDIA AI Factories Globally for Drug Discovery

Roche is deploying over 3,500 NVIDIA Blackwell GPUs worldwide to accelerate drug discovery, diagnostic solutions, and manufacturing. This large-scale integration across its value chain aims to massively boost R&D productivity, enable next-generation diagnostics, and improve manufacturing efficiency through advanced AI infrastructure.

2026-03-16

Musk's xAI Restarts Effort to Build AI Coding Tool

Elon Musk's xAI is revamping its effort to build an AI coding tool, with two new executives joining from Cursor. The restart suggests previous attempts were 'not built right the first time,' highlighting the challenges in developing effective AI-powered software engineering assistants in a competitive market.

2026-03-16

Balyasny Asset Management Builds AI Research Engine with GPT-5.4

Balyasny Asset Management has built an AI research system for investing using GPT-5.4, rigorous model evaluation, and agent workflows. The engine transforms investment analysis at scale, demonstrating how advanced AI models are being integrated into high-stakes financial decision-making processes to parse data and generate insights.

Palantir Demos AI Chatbots for Military War Planning

Palantir software demos and Pentagon records detail how AI chatbots like Anthropic's Claude could analyze intelligence and suggest war plans. This showcases the potential military applications of generative AI for strategic planning, raising questions about the role of automation in defense decision-making.

Physical AI Emerges as Key Advantage for Manufacturing

Physical AI is becoming manufacturing's next competitive advantage, moving beyond traditional automation. It helps manufacturers grow amid labor constraints and rising complexity by integrating AI with physical processes, enabling adaptive, intelligent systems that optimize production in real-time.

Lawyer Warns of AI Chatbot Risks in Mass Casualty Cases

A lawyer involved in AI psychosis cases warns that AI chatbots are now appearing in mass casualty incidents, not just suicides. The warning highlights that AI technology is advancing faster than safeguards, raising urgent concerns about liability and safety regulations for generative AI systems.

Meta Considers Layoffs Affecting 20% of Staff

Meta is reportedly considering layoffs that could affect up to 20% of its workforce, potentially eliminating around 15,800 positions. Sources indicate the move aims to offset aggressive spending on AI infrastructure, acquisitions, and hiring, reflecting the high cost of competing in the AI race.

Open Source

2026-03-14

China's OpenClaw Boom Drives AI Company Revenue

Hype around the open-source AI agent OpenClaw is driving significant commercial activity in China, with users spending on cloud servers and AI subscriptions. This trend is creating substantial revenue opportunities for tech companies providing the necessary infrastructure. The rapid adoption illustrates how open-source AI tools can catalyze market growth and innovation.

2026-03-14

Microsoft's Copilot AI Assistant Coming to Xbox Consoles

Microsoft will launch its Gaming Copilot AI assistant on current-generation Xbox consoles this year. The AI feature was announced at the Game Developers Conference and represents Microsoft's expansion of Copilot into gaming platforms. This integration aims to provide players with in-game assistance, tutorials, and enhanced interactive experiences through AI.

2026-03-14

Why Physical AI Is Manufacturing's Next Competitive Edge

Manufacturers are adopting physical AI to overcome labor constraints, rising complexity, and supply chain volatility. This approach goes beyond traditional automation by integrating AI with physical systems to optimize design, simulation, and operations. The technology enables more adaptive and efficient manufacturing processes, representing the next phase of industrial advancement.

Manufact Raises $6.3M as MCP Becomes 'USB-C for AI'

Startup Manufact raised $6.3 million as its Model Context Protocol (MCP) gains traction as a universal standard for connecting AI applications to data sources and tools. Dubbed the 'USB-C for AI,' MCP is now powering integrations for ChatGPT and Claude, enabling seamless communication between agents and external systems. This reflects a shift towards standardized interfaces in the fragmented AI tooling landscape.

Anthropic's Claude Gains Shared Context Across Excel & PowerPoint

Anthropic has upgraded Claude with new capabilities for Microsoft Excel and PowerPoint, enabling shared context and reusable workflows across the applications. This allows the AI to maintain consistency and build upon previous work within Office documents. The move expands Claude's enterprise utility and positions it as a potential competitor to Microsoft's own Copilot offerings, which Claude also powers.

NVIDIA's Nemotron 3 Super Combines Architectures for Efficiency

NVIDIA's new open-weights Nemotron 3 Super model uniquely combines three different neural architectures to outperform other open models in multi-agent tasks. It is designed to handle the high token volume generated by agentic systems (up to 15x standard chats) cost-effectively. This innovation addresses a key bottleneck for deploying complex, long-horizon AI agents in enterprise environments.

AI Error Jails Innocent Woman in Facial Recognition Case

An innocent grandmother was jailed for months in North Dakota after being misidentified by an AI facial recognition system in a fraud case. The incident underscores the serious real-world consequences of AI errors in law enforcement and justice systems. It has sparked renewed debate about the reliability, oversight, and ethical use of biometric AI technologies in criminal investigations.

Document Poisoning Threatens RAG System Security

A new report details how attackers can corrupt Retrieval-Augmented Generation (RAG) systems by poisoning the documents they use as sources. By injecting malicious or misleading data into source materials, bad actors can manipulate AI outputs without directly attacking the model. This vulnerability highlights emerging security challenges as enterprises deploy RAG for knowledge-intensive applications.

Multimodal

Anthropic's Claude AI Can Now Generate Charts and Diagrams

Anthropic has updated its Claude AI to generate custom charts, diagrams, and visualizations directly within conversations. The AI determines when a visual aid is useful based on context and inserts images in-line rather than in a side panel. This multimodal enhancement aims to make complex information more accessible and improve user comprehension during technical or data-heavy discussions.

US Military Explores AI Chatbots for Target Recommendation

A US Defense Department official revealed that generative AI systems could be used to rank target lists and make strike recommendations, which would then be vetted by humans. This potential use of AI chatbots in military decision-making comes amid ongoing ethical and operational debates about autonomous weapons systems. The disclosure highlights the increasing integration of AI into defense and targeting workflows.

Meta Develops Four New Chips for AI and Recommendations

Meta is developing four new versions of its MTIA (Meta Training and Inference Accelerator) chips to power its AI and recommendation systems. This move represents the tech giant's latest effort to build its own custom AI hardware, aiming to reduce reliance and costs associated with chips from industry leaders like Nvidia, while optimizing for its specific workloads.

Nvidia to Spend $26B on Open-Weight AI Models

Nvidia plans to invest $26 billion in building open-weight AI models, according to filings. This massive investment could quickly position the chipmaker and infrastructure leader as a direct competitor to model companies like OpenAI and Anthropic, expanding its influence from hardware into the foundational model layer of the AI stack.

Multimodal

OpenAI's Sora Video Generator May Integrate into ChatGPT

OpenAI's Sora video generation model could soon become a built-in feature within ChatGPT, according to a report. Currently a standalone product, integrating Sora into the popular chatbot would significantly broaden access and simplify the video creation workflow for users, potentially boosting its adoption and utility.

Open Source

Hustlers Cash In on China's OpenClaw AI Craze

A gold rush is emerging around OpenClaw, a popular open-source AI tool in China that can autonomously control devices. Entrepreneurs and developers are quickly building businesses and services on top of the tool, capitalizing on its capabilities for automation. The trend highlights the rapid commercialization of powerful, accessible AI agent frameworks.

Meta's Moltbook Deal Signals Future Built on AI Agents

Meta's acquisition of Moltbook may signal its vision for a future shaped by AI agents, particularly in advertising and commerce. The deal suggests Meta is investing in infrastructure for an 'agentic web,' where autonomous AI assistants handle tasks like shopping and content discovery, potentially reshaping user interaction with its platforms.

Replit Valuation Hits $9B, Up from $3B Six Months Ago

AI-powered development platform Replit has raised a new $400 million round, reaching a $9 billion valuation just six months after hitting $3 billion. The company stated it hopes to achieve $1 billion in annual recurring revenue by year's end. This rapid growth underscores the high demand for cloud-based, AI-assisted coding environments.

Zendesk Acquires Agentic AI Startup Forethought

Zendesk has acquired Forethought, an agentic customer service AI startup that was the 2018 winner of TechCrunch Battlefield. The acquisition signals Zendesk's push to integrate more advanced, autonomous AI agents into its customer service platform, moving beyond simple chatbots to systems that can handle complex, multi-step support workflows.

OpenAI Details Agent Runtime with Computer Environment

OpenAI explains how it built a secure, scalable agent runtime using the Responses API, a shell tool, and hosted containers. This system equips agents with a computer environment, allowing them to use files, tools, and maintain state. The development addresses key challenges in creating reliable and safe autonomous AI agents for complex workflows.

Rakuten Cuts Fix Time 50% Using OpenAI's Codex

Rakuten uses OpenAI's Codex coding agent to ship software faster and safer, reducing its Mean Time to Resolution (MTTR) by 50%. The tool automates CI/CD reviews and enables the delivery of full-stack builds in weeks instead of months. This case study demonstrates the significant impact of AI coding assistants on enterprise development velocity and operational efficiency.

NVIDIA Nemotron 3 Super Boosts Agentic AI Throughput 5x

NVIDIA has launched Nemotron 3 Super, a 120-billion-parameter open model with 12 billion active parameters designed for scalable agentic AI systems. The model delivers five times higher throughput and combines advanced reasoning to complete complex tasks with high accuracy for autonomous agents. It is available now, aiming to make large-scale multi-agent workflows more cost-effective and efficient.

Open Source

RunAnywhere Launches Tool for Faster AI Inference on Apple Silicon

RunAnywhere (YC W26) has launched an open-source tool called 'rcli' designed to significantly accelerate AI inference on Apple Silicon chips. The project, featured on Hacker News, aims to optimize performance for developers running AI models locally on Macs, improving speed and efficiency.

Judge Blocks Perplexity AI Agents from Shopping on Amazon

A US federal judge has issued an order blocking Perplexity's web browser-based AI agents from placing orders on Amazon on a user's behalf. The judge found Amazon provided strong evidence that Perplexity's 'Comet' browser agents violated its terms of service, highlighting legal friction around autonomous AI agents.

Multimodal

YouTube Expands AI Deepfake Detection to Public Figures

YouTube is expanding access to its AI deepfake detection tool, making it available to politicians, government officials, and journalists. This allows these public figures to flag unauthorized uses of their likeness in videos for potential removal, addressing the growing challenge of AI-generated misinformation.

Ford Gives Commercial Fleet Business an AI Makeover

Ford announced 'Ford Pro AI,' a new generative AI service for its commercial fleet and telematics software. The system analyzes vehicle data—like speed, seat belt use, and engine health—and converts it into actionable insights for fleet managers, aiming to improve safety, efficiency, and maintenance.

OpenAI's Codex Security Enters Research Preview

OpenAI's Codex Security, an AI application security agent, is now available in a research preview. It analyzes project context to detect, validate, and patch complex code vulnerabilities with higher confidence and less noise than traditional tools, representing a new approach to automated security.

Amazon Launches Healthcare AI Assistant on Website and App

Amazon has launched a healthcare AI assistant directly on its website and mobile app. The assistant can answer health-related questions, explain medical records, manage prescription renewals, and help book appointments, expanding Amazon's integration of AI into consumer health services.

AI Art

NVIDIA & ComfyUI Streamline Local AI Video Generation at GDC

At the Game Developers Conference, NVIDIA announced updates that streamline local AI video generation for creators using tools like ComfyUI. The enhancements, showcased on RTX AI PCs, are designed to help game developers and artists more efficiently build cinematic worlds and characters during the concept development phase.

OpenAI Improves LLM Safety with Instruction Hierarchy Challenge

OpenAI has developed the Instruction Hierarchy Challenge (IH-Challenge) to train frontier LLMs to better prioritize trusted instructions. This improves 'instruction hierarchy,' enhances safety steerability, and increases model resistance to prompt injection attacks, addressing key security concerns in advanced AI systems.

Anthropic Sues US Defense Dept Over 'Supply-Chain Risk' Label

AI startup Anthropic is suing the US Department of Defense, challenging the Trump administration's decision to designate the company a 'supply-chain risk.' Anthropic argues the move oversteps legal authority and has escalated a contract dispute into an effective federal ban, potentially costing the company billions in revenue as partners pause deals.

Yann LeCun Raises $1B for AI That Understands Physical World

Meta's former chief AI scientist, Yann LeCun, has raised $1 billion for his new startup, AMI. The company's goal is to build AI that masters understanding of the physical world, a departure from the language-focused approach of current LLMs. LeCun has long argued that human-level AI requires this foundational grasp of physics and reality.

NVIDIA & Thinking Machines Lab Form Gigawatt-Scale AI Partnership

NVIDIA and Thinking Machines Lab announced a multi-year strategic partnership to deploy at least one gigawatt of next-generation NVIDIA Vera Rubin systems. This massive compute power will support the training of frontier AI models and platforms designed to deliver customizable AI at an unprecedented scale, marking a major infrastructure investment.

Microsoft Warns Ungoverned AI Agents Could Be 'Double Agents'

Microsoft warns that ungoverned AI agents operating within corporations could become 'double agents.' In response, it announces the general availability of Agent 365 and Microsoft 365 Enterprise 7, products designed to bring security and governance to organizational AI agents. The tools aim to manage the risks associated with deploying autonomous AI at scale.

Microsoft Announces Copilot Cowork with Anthropic's Help

Microsoft announces 'Copilot Cowork,' a new cloud-powered AI agent within Microsoft 365 Copilot developed with help from Anthropic. The agent can complete work across multiple applications autonomously. This partnership highlights the deepening integration of frontier AI models into mainstream enterprise productivity suites to enable complex, cross-app workflows.

Anthropic Rolls Out Code Review Amid Pentagon Legal Battle

Anthropic releases Code Review for Claude Code, a multi-agent system that scrutinizes pull requests for bugs, even as the company sues the Pentagon over a supply-chain risk designation. The tool is available in research preview for Team and Enterprise customers, aiming to improve code quality and security in AI-augmented development environments.