All Updates
Page 1092 of 1566
March 25, 2026
Hitachi's LogiRiSM AI Boosts Logistics 4x
Hitachi announced LogiRiSM, a transport plan optimization AI engine that evolves logistics center material handling equipment into physical AI. It monitors equipment states in real-time, achieving about 4x productivity in sorting tasks.
SK Hynix: AI Drives Long-term Memory Boom
SK Hynix states that AI will expand memory demand in the medium to long term. HBM is expected to remain a key growth driver in the memory market through 2026.
Nvidia L4 Strategy Misunderstood
Nvidia's L4 strategy has been widely misunderstood. The company has been deeply invested in autonomous driving for over a decade. This analysis corrects misconceptions about their long-term approach.
Openreach Scales Google AI for Fiber Fleet
Openreach expands Google Cloud partnership using AI to speed fiber broadband rollout and cut emissions. AI analyzes routes, idling, and faults for its 24,000-truck fleet traveling 200M miles yearly.
OpenAI Foundation Unveils $25B AI Research Roadmap
OpenAI Foundation announced a $25B investment in AI research, starting with $1B next year, post-recapitalization where it holds equity in for-profit OpenAI. Focus areas include AI resiliency in biosecurity, model safety, and impacts on children/youth, plus life sciences for disease research. This roadmap aims to build safeguards, shared data ecosystems, and serve as a blueprint for AI development.
TikTok Launches US Self-Made Short Dramas
TikTok is recruiting actors for its first self-produced short drama series in the US, styled as a soap opera. The platform filed a 'TikTok Drama' trademark last year and currently features AI-enhanced third-party short dramas. This move comes amid a booming $1.4B US micro-short drama market.
Safari 26.4 Restores Compact Tabs, Adds Web APIs
Safari 26.4 fixes 191 bugs and adds 44 features, including compact tab layout revival on macOS. Key additions: CSS Grid Lanes for layouts, WebTransport for low-latency, and Keyboard Lock API for web apps. Deprecated FontFaceSet constructor removed.
Shenzhen Pushes 1.6T Optical Modules
Shenzhen's action plan accelerates AI server chain by upgrading optical modules from 800G to 1.6T/3.2T. Supports mass production of 800G+ modules and advances silicon photonics, CPO/LPO. Boosts core tech like thin-film lithium niobate and InP.
AstraZeneca-Tsinghua Launch AI Drug Center
AstraZeneca and Tsinghua University signed a cooperation agreement to establish the Tsinghua AIR-AstraZeneca AI Drug R&D Joint Center. The center focuses on AI-driven molecule research, translational medicine, and clinical innovation. It aims to integrate AI agents across the full drug development pipeline, shifting from single-step to global decision-making.
AI Drives Efficiency via Unemployment
UK data reveals rising unemployment alongside productivity gains, mirroring France's capital-deepening model where high labor costs spur automation. AI accelerates this labor substitution, potentially displacing more cognitive jobs than it creates. Article explores UBI and redefining work's meaning in AI era.
Real World Needs Algorithm 'Resistance'
Philosopher Wang Xiaowei argues for 'resistance' against pervasive algorithms in life to reclaim subjective experiences. He critiques info cocoons, loss of judgment sovereignty, and high-tech's monitoring effects. Advocates low-tech routines like market visits for genuine connections.
HP IQ Integrates OpenAI LLM in Laptops
HP launches HP IQ, a local AI collaboration app powered by OpenAI LLM for its business laptops. It enables chatting, file sharing, meeting recording, and summarization. This positions HP against rivals like Apple Intelligence, Google Gemini, and Microsoft Copilot.
Australia's Narrow AI Hub Window
Australia has a narrow window to become a regional AI infrastructure hub. It risks falling behind without accelerated investment and policy action. This assessment comes from Deloitte Access Economics.
Meta Glasses Delayed in EU
Meta's new display-equipped Ray-Ban smart glasses rollout is stalled in the EU. Issues include battery regulations, AI rules, and supply shortages. This hampers European market entry.
xMemory Cuts Token Costs in AI Agents
xMemory organizes conversations into a searchable hierarchy of semantic themes to solve RAG limitations in long-term AI agent deployments. It improves answer quality, long-range reasoning, and cuts token usage from over 9,000 to 4,700 per query. This enables cost-effective, coherent multi-session AI assistants for enterprises.
OpenAI Safety Bug Bounty Launch
OpenAI launches a Safety Bug Bounty program to crowdsource AI risk detection. It focuses on abuse cases like agentic vulnerabilities, prompt injection, and data exfiltration.
Autodesk Fusion Speeds Para-Athlete Prosthetics
Autodesk partnered with BioDapt to accelerate next-gen high-performance prosthetics for para-athletes. They integrated accumulated development data and existing CAD models into Autodesk Fusion. A 'Fusion Hub' was built for unified cloud-based management.
Apple's XSA Boosts Transformer Performance
Apple introduces exclusive self attention (XSA), a simple tweak to self attention that excludes the token's own value vector for better context modeling. XSA outperforms standard self attention across models up to 2.7B parameters on language modeling tasks. Gains grow larger with increasing sequence lengths.
Apple's Latent Lookahead for Transformers
Apple's paper on Latent Lookahead Training for Transformers was accepted at the ICLR 2026 workshop on Latent & Implicit Thinking. It tackles autoregressive models' limitations by enabling exploration of multiple continuations without early commitment. This allows non-uniform compute allocation to boost expressiveness on challenging tokens.