NerdNews

Elon Musk's AI company xAI has acquired Hotshot, a startup that developed AI-powered video generation tools. Hotshot's technology will be integrated into xAI's Grok chatbot platform, which is expected to release a 'Grok Video' model in the coming months. The acquisition indicates xAI's plans to compete with other video generation models like OpenAI's Sora and Google's Veo 2.

OpenAI is launching a beta test for ChatGPT Connectors, allowing businesses to link Google Drive and Slack accounts to ChatGPT, enabling the chatbot to answer questions based on internal company information. The feature will expand to other platforms like Microsoft SharePoint and Box in the future.

Google announced the addition of Chirp 3, its speech-to-text and HD text-to-speech models, to its Vertex AI development platform. This move is part of the company's efforts to expand its AI capabilities and provide developers with more tools to build voice-based applications. Chirp 3 will be available on Vertex AI starting next week, with use cases including building voice assistants, creating audiobooks, and developing support agents and voice-overs for videos.

Google is replacing its voice Assistant with the Gemini AI service. The change will occur over the coming months, with the classic Google Assistant no longer accessible on most mobile devices later this year. Gemini will also be integrated into home devices like speakers, displays, and TVs, offering a new experience.

Google's new Gemini Flash 2.0 image generation tool can remove watermarks from copyrighted images. The model is currently in its experimental phase and is only available to developers. While it's not perfect, it has been shown to be effective in removing watermarks, which could have significant implications for copyright and intellectual property protection.

Meta has unified Facebook's video delivery system across mobile and web apps by migrating video experiences from Facebook Watch to Reels, resulting in a seamless transition for billions of users. The unification process involved changes to mobile UI, server backend, and ranking systems, and included improvements such as dynamic pagination and real-time ranking.

Moumita Bhattacharya discusses large-scale recommender and search systems, highlighting common components, data requirements, and infrastructural needs. She shares examples from Netflix, including a pre-query recommendation model and a unified contextual recommender system, UniCoRn, which combines search and recommendation tasks into one model.

This article explores efficient preloading systems for mobile video streaming apps, balancing user experience with technical constraints. It covers key technical strategies such as dynamic buffering, adaptive playback, and edge computing to ensure seamless playback and efficient content delivery.

Felix GV discusses how to create and operate planet-scale data storage solutions for derived data, including the use of Kafka, Apache Helix, and RocksDB, and the importance of loose coupling and optimization for performance

The .NET Team has announced the release of HybridCache, a new .NET 9 library that combines in-memory and distributed caches to enhance data storage and retrieval performance. The library simplifies common development tasks, improves application speed and responsiveness, and provides features such as cache-stampede protection and cache invalidation using tags.

Nvidia's GTC 2025 conference will feature a keynote by CEO Jensen Huang, with expected announcements on Blackwell Ultra GPU series, Rubin chip architecture, and AI updates. The conference will be livestreamed, and sessions will be available on-demand, including workshops on large language model customization and generative AI for core banking.

Palmetto is opening its AI building models to outside developers to speed up the electrification of US building stock. The models use public and private data to simulate digital twins of every residential building in the US, allowing for more accurate assessments of electrification projects such as solar panels and heat pumps.

Roblox has released an open-source model called Cube that can create 3D objects using generative AI. The model allows creators to generate 3D meshes using a single prompt and can be customized and trained with user datasets. Roblox also announced three additional AI tools: text generation, text-to-speech, and speech-to-text, which will launch in the coming months.

Discord has launched a free Social SDK to help game developers enhance social experiences in their games. The toolkit allows for unified friends lists, cross-platform messaging, and voice chat. It is compatible with C++, Unreal Engine, and Unity, and supports Windows 11+ and macOS, with console and mobile support coming soon.

Cloudflare has introduced end-to-end post-quantum cryptography protections in its Zero Trust platform, allowing organizations to safeguard their corporate network traffic from potential quantum computing attacks without individually upgrading each application or system. This move aligns with NIST's efforts to transition away from conventional cryptographic algorithms.

Google's Gemini Robotics AI model combines language, vision, and physical action to power capable robots. The model, announced by Google DeepMind, can control different robots in various scenarios and has the potential to revolutionize robotics. Google also introduced a benchmark called ASIMOV to gauge risks with AI-powered robots.

Google's parent company Alphabet is in advanced talks to acquire cloud cybersecurity startup Wiz at a valuation of around $30 billion, higher than the $23 billion discussed last summer. The deal, led by Google Cloud's Thomas Kurian, aims to integrate Wiz's cloud security products into Google Cloud, attracting regulatory scrutiny.

Nvidia's GTC 2025 conference is expected to feature announcements on AI, GPUs, and quantum computing. CEO Jensen Huang will give a keynote address on Tuesday, focusing on AI and accelerating computing technologies. The company may reveal new GPU iterations, including Blackwell Ultra and Rubin, and discuss its approach to quantum computing.

Adobe data shows AI search is on the rise, with referrals increasing 1,300% during the 2024 holiday season. Users who come from AI search tend to stay on sites longer and browse more pages. Google, Perplexity, and OpenAI are major players in the AI search space, with varying approaches to advertising and content generation.

Google is transitioning to Gemini and removing popular Assistant features, including photo sharing, interpreter mode, and daily updates. Gemini will replace Assistant later this year, but may lack direct replacements for some features.