AI News
500 articles — Page 27 of 42
IIT Kanpur startups unveil kamikaze drone and cyber defense tools
At IIT Kanpur's C3iHub Startup Demo Day Connect 1.0, five incubated ventures debuted drones, blockchain forensics, and cyber-threat platforms. The showcases highlight India's push toward technological self-reliance and closer government-academia-private sector collaboration to strengthen national security and digital infrastructure. The launches signal a shift from R&D to market-ready solutions across surveillance, cybersecurity, and forensics.
IBM and NASA unveil Surya AI to predict solar flares hours ahead
IBM and NASA have introduced Surya, an open-source AI model that forecasts solar flares up to two hours ahead and pins down where on the sun’s surface an eruption might originate. Built on nine years of SDO data and released on Hugging Face as SuryaBench, early tests show meaningful accuracy gains and broader access to help protect satellites, power grids, and aviation.
Zed Bets $32M on DeltaDB to Disrupt Git
Zed, an AI coding startup, has raised $32 million in a Series B led by Sequoia Capital to advance DeltaDB, a novel operation-based version control system. DeltaDB aims to fuse real-time code edits with the surrounding conversations between humans and AI agents, challenging Git's snapshot-based model and signaling a broader shift toward collaborative, AI-powered development tools.
AI Turns Imagination Into Playable Worlds
Dynamics Lab's Mirage 2 turns simple images into navigable 3D spaces in real time. Users can shape environments with text prompts, save and share worlds, all streamed from the cloud for in-browser play. It hints at a future where game creation is more accessible and collaborative.
Wipro to acquire Harman DTS for $375M to boost AI-led engineering
Wipro announced a definitive agreement to purchase Harman's Digital Transformation Solutions unit for up to $375 million in cash. The deal, expected to close by December 31, 2025, includes more than 5,600 DTS employees and seeks to strengthen Wipro's AI-powered digital engineering capabilities, global delivery footprint, and strategic collaboration with Harman and Samsung.
OpenAI opens New Delhi office, deepening India push
OpenAI confirms its first office in India, establishing a New Delhi base and initiating local hiring. The move is part of a broader strategy that includes localized pricing, payment integrations, and government collaboration to broaden access to its AI tools for India's vast user and developer communities. The expansion underscores India's growing importance on the global AI map.
Cadence and NVIDIA reveal 97% pre-silicon AI chip power modeling
Cadence Design Systems and NVIDIA unveil the Dynamic Power Analysis app on the Palladium Z3 platform, delivering up to 97% accuracy in predicting AI chip power before fabrication. The tool aims to accelerate design cycles and boost energy efficiency by validating power profiles with real workloads early in development.
ElevenLabs v3 adds emotional depth and multi-speaker dialogue to TTS
ElevenLabs released an alpha of Eleven v3, a text-to-speech model centered on emotional nuance, context awareness, and multi-speaker dialogue from a single script. It introduces audio tags for directorial control, improved pacing, and broad language support across over 70 languages. The alpha signals a potential shift for creators, while also highlighting early tradeoffs around cloning fidelity and latency.
Google Claims Gemini Prompt Uses Less Energy Than 9 Seconds of TV
Google released figures asserting that a median Gemini text prompt uses 0.24 watt-hours of energy, emits 0.03 g CO2e, and consumes roughly five drops of water, framed within a full-stack efficiency approach. Critics argue the numbers may overlook indirect water use, energy-grid mix, and lifecycle costs, complicating the claimed environmental advantage. The piece reviews the methodology and broader implications for AI sustainability.
Confluent Launches Streaming Agents to Bridge Real-Time Data and AI
Confluent today introduced Streaming Agents, a feature for Confluent Cloud built on Apache Flink that threads live data streams directly into agentic AI workflows. The update aims to move AI projects from prototyping to production by unifying data processing with AI reasoning and tool calling, enabling context-aware automation across business processes. It also includes replayability for safe testing with historical data.
Nvidia Develops B30A for China Amid U.S. Rules
Nvidia is developing a more powerful Blackwell-based chip for the Chinese market, the B30A, to sustain its Chinese revenue while staying within U.S. export controls. The project signals a continued push to compete with domestic rivals like Huawei while navigating a tightening regulatory landscape, and includes plans for samples and a companion RTX6000D for lighter tasks.
GitHub Unveils Agents Panel to Orchestrate AI Coding
GitHub introduces the Agents Panel, a centralized cockpit to manage Copilot-powered AI agents handling coding tasks. The feature aims to reduce context switching, enable asynchronous work, and elevate developer workflows by letting humans direct autonomous agents from within the platform. Public preview is available to paid Copilot subscribers.