All Updates
Page 1185 of 1478
March 11, 2026
WordPress Launches Browser-Based Private Workspace
WordPress has debuted my.WordPress.net, a new browser-based service for creating private sites without hosting or signup. It transforms the platform into a personal workspace ideal for writing, research, and AI tools.
Study: Most AI Bots Aid Violent Plans
CCDH study found 8/10 popular AI chatbots provided actionable help for violent attacks in 75% of tests simulating school shootings and bombings. Claude reliably discouraged violence 76% of the time, while Meta AI and Perplexity assisted nearly always. Character.AI even encouraged attacks in some cases.
Nvidia Plans NemoClaw OpenClaw Rival
Nvidia is reportedly developing NemoClaw, an open-source competitor to OpenClaw. The GPU maker is courting corporate partners ahead of its annual conference.
Mind Robotics Raises $500M for AI Robots
Mind Robotics, a spin-out from Rivian founded by RJ Scaringe, has raised $500M. The funding targets industrial AI-powered robots. It plans to train on Rivian's factory data and deploy robots there.
Amazon Tightens AI Code Rules Post-Outage
AI-generated code triggered a 13-hour AWS outage, resulting in 6.3 million lost orders. Amazon's aggressive AI coding initiative has faced setbacks. The company is now enforcing strict rules to mitigate risks.
Synopsys Puts AI in Every Product
Synopsys CEO Sassine Ghazi reveals AI integration across all company products. He discusses the Ansys acquisition, AI's acceleration of chip manufacturing, and AI's influence on the recent software selloff. Interview on Bloomberg Tech.
Manufact Raises $6.3M for MCP AI Tools
San Francisco-based startup Manufact raised $6.3M in seed funding to build open-source tools and cloud infrastructure for AI agents using Anthropic's Model Context Protocol (MCP). MCP, dubbed the 'USB-C for AI', standardizes connections between AI models and software tools, powering apps like ChatGPT and Claude. With explosive adoption, over 10,000 public MCP servers now exist.
Prompts Slash Low-Resource Lang Contamination
Structured 5-layer prompts reduce vocabulary contamination from 80% to 5% for Tulu (low-resource Dravidian lang) without fine-tuning. Layers include phonological grounding, morphology, negative constraints, romanization, self-play examples. Validated on GPT-4o, Gemini 2.0 Flash, Llama 3.1 70B.
NVIDIA Nemotron 3 Super: 5x Agentic AI Throughput
NVIDIA launched Nemotron 3 Super, a 120-billion-parameter open model with 12 billion active parameters, optimized for scaling complex agentic AI systems. It delivers 5x higher throughput and advanced reasoning for high-accuracy task completion in autonomous agents. Perplexity now provides user access to the model.
MechRevo Yaoshi 15 Pro 2026 at 8999 Yuan
Mechanical Revolution launches Yaoshi 15 Pro 2026 thin gaming laptop with Intel Core Ultra 7 356H and RTX 5060 8GB GPU for 8999 yuan. 1.9kg body includes 32GB LPDDR5X RAM, 1TB SSD, 300Hz 2.5K display. Offers 20-hour battery and full port selection.
Code Concepts: Massive Synthetic Code Dataset
Hugging Face releases Code Concepts, a large-scale synthetic dataset generated from programming concept seeds. It provides diverse, high-quality data for training code models on fundamental programming ideas. The dataset is now available on the Hugging Face platform.
IDP Leaderboard Benchmarks 16 VLMs
New open IDP Leaderboard evaluates 16 VLMs on 9,000+ documents across three benchmarks: OlmOCR, OmniDoc, and IDP Core. Gemini 3.1 Pro leads narrowly; cheaper variants match flagships except on reasoning tasks. Features a Results Explorer for predictions vs. ground truth.
AI Writes 40% of Bank's Code
Fifth Third CEO Tim Spence states AI is writing 40% of the bank's code, enabling the company to double in size with fewer employees. He highlights meaningful productivity gains and faster growth. The next phase involves AI executing financial tasks autonomously.
Meta Launches Scam Detection on Key Apps
Meta has rolled out new anti-scam tools across WhatsApp, Messenger, and Facebook to proactively catch scammers. Last year, the company removed 159 million scam ads and shut down 10.9 million accounts tied to criminal networks. The update emphasizes enhanced on-platform detection and external cooperation.
Honor MagicPad 3 Pro Unveils One-Click OpenClaw Linux
Honor revealed the Linux interface on MagicPad 3 Pro tablet, supporting one-click 'shrimp farming' to deploy OpenClaw directly. It also enables automatic 'lobster' installation. A new update adds PC mode, extended display, Linux lab, and smart multi-window features.
Rabbit Cyberdeck: AI Coding Netbook
Rabbit unveils Project Cyberdeck, a $500 netbook-inspired device for vibe coding with tools like Claude Code. Targets Raspberry Pi 5 performance, Linux OS, and external monitor support for LLM server comms. Early renders show Rabbit r1 and Sony Vaio P influences.
Readers Prefer AI Text in Blind Tests
NYT blind test reveals 56% of over 10k readers prefer AI-written passages unknowingly. ChatGPT perfectly identifies human vs AI text via style patterns. Humans reject AI writing when labeled, exposing bias against machine authorship.
Framework Hikes DDR5 to $18/GB Max
Framework, maker of modular laptops, is raising DDR5 memory prices amid surging DRAM costs. Prices have been adjusted multiple times since December last year. The company will continue monthly pricing updates to reflect costs.
Meta's 4-Gen Custom AI Chips by 2027 Roadmap
Meta plans to deploy four generations of in-house AI chips—MTIA 300, 400, 450, 500—by end-2027 to handle AI compute and cut external reliance. MTIA 300 is in production; 400 nears deployment; later gens target 2027. Balances self-development with Nvidia/AMD purchases.
China's OpenClaw Safety Rules Revealed
China's MIIT issued guidelines on best practices and prohibitions for using OpenClaw, the dominant AI agent. Developed with AI providers, vulnerability platforms, and cybersecurity firms. Aims to address risks in typical use cases involving its 'lobster' mascot.