AI News Recap: May 30, 2025
Tech Giants Race to Integrate AI: Meta, OpenAI, and Google Unveil New Features
So, What’s Happening in AI This Week?
Odyssey’s building AI worlds you can explore in real time, and Meta’s splitting its AI department to chase faster product launches. DeepSeek’s new model is huge and open for business, but the U.S. isn’t happy about it. The New York Times did a deal with Amazon to share content for AI, and Google Photos’ latest AI editing tools are rolling out. Anthropic’s giving Claude a voice, and DeepSeek’s smaller models are making big waves on benchmarks.
👋 Catch up on the latest post …
In the Spotlight 🔦
Odyssey Unveils AI Model for Streaming Interactive 3D Worlds
Category: Emerging Technologies
🕹️ Odyssey's new AI model generates and streams interactive 3D video worlds in real time, allowing users to explore environments like a video game.
🚀 The model streams video frames every 40 milliseconds and can maintain coherent output for over 5 minutes, currently running at up to 30 FPS using Nvidia H100 GPUs.
🎬 Odyssey aims to collaborate with creatives and is developing tools for integration with platforms like Unreal Engine, addressing concerns about job disruption in entertainment industries.
Leading European AI Researcher Raises $13M to Pursue Breakthrough Model
Category: Research & Development
🚀 SpAItial, founded by AI researcher Matthias Niessner, raised a $13M seed round to develop text-to-3D foundation models that generate interactive, realistic environments.
🧑💻 The startup's team includes ex-Google and Meta experts in 3D and AI, aiming to achieve the 'Holy Grail' of fully interactable, photorealistic 3D worlds from text prompts.
🎮 Potential applications span video games, entertainment, construction visualization, and robotics, with SpAItial prioritizing revenue generation and developer partnerships for early adoption.
Meta Divides AI Team to Accelerate Product Development
Category: Business & Industry Trends
🚀 Meta is splitting its AI department into two teams: AI Products and AGI Foundations, aiming to accelerate product development.
📱 The AI Products team will focus on consumer-facing features in Facebook, Instagram, WhatsApp, and Meta’s standalone AI app.
🤖 The AGI Foundations unit will work on foundational projects like enhancing Meta’s Llama models to stay competitive with OpenAI and Google.
OpenAI Plans to Introduce 'Sign In with ChatGPT' Feature for Third-Party Apps
Category: Applications
🔑 OpenAI is exploring a 'Sign in with ChatGPT' feature for third-party apps, aiming to simplify user authentication across various platforms.
📈 ChatGPT has around 600 million monthly active users, positioning OpenAI to compete with tech giants like Apple, Google, and Microsoft in consumer services.
🛠️ A developer preview of the feature launched in Codex CLI, offering API credits to users who sign in with ChatGPT, and gauging interest from companies of all sizes.
Anthropic Introduces New Voice Mode for Claude AI
Category: Applications
🗣️ Anthropic launched a beta voice mode for Claude, enabling users to have spoken conversations and receive voice responses in English on the mobile app.
🔄 Users can switch between text and voice, choose from five voices, and see on-screen transcripts and summaries during conversations with Claude.
🔒 Voice mode features are limited by subscription: free users have capped conversations, while paid subscribers get Google Workspace integration for calendar and email access.
Google Photos Unveils Redesigned Editor Featuring Advanced AI Tools
Category: Applications
🖼️ Google Photos introduces a redesigned editor featuring new AI tools—Reimagine and Auto Frame—previously exclusive to Pixel devices, now available to a broader user base.
✨ The Reimagine feature uses generative AI to transform photo objects and backgrounds via text prompts, enabling creative edits like changing skies or backgrounds.
📱 The update rolls out globally to Android devices next month, with iOS support coming later this year, expanding access to AI-powered photo editing.
DeepSeek Releases Updated R1 Reasoning AI Model on Hugging Face
Category: Research & Development
🚀 DeepSeek released an updated version of its R1 reasoning AI model on Hugging Face, making it available under a permissive MIT license for commercial use.
🧠 The updated R1 model is substantial, with 685 billion parameters, making it too large to run on most consumer-grade hardware without modification.
🌏 DeepSeek has attracted regulatory attention in the US, with concerns raised about its technology posing potential national security risks.
DeepSeek: Key Features and Insights About the AI Chatbot App
Category: Business & Industry Trends
🚀 DeepSeek's AI chatbot app quickly topped app store charts, challenging U.S. dominance and prompting questions about the future of global AI competition.
💡 DeepSeek's models like V3 and R1 outperform many rivals in benchmarks, drive down prices, and are widely adopted by developers under permissive commercial licenses.
⚠️ DeepSeek faces bans from U.S. government devices and other countries due to concerns over data security, propaganda, and regulatory compliance with Chinese government values.
New York Times Signs AI Licensing Agreement With Amazon
Category: Business & Industry Trends
🤝 The New York Times has signed its first AI content licensing deal with Amazon to provide editorial material for AI training and customer experiences.
📰 The agreement covers news articles, NYT Cooking, and The Athletic, with potential integration into Amazon Alexa and other platforms, enhancing content accessibility.
💡 This marks The Times' first generative AI-focused licensing deal, following its previous copyright lawsuit against OpenAI and Microsoft for unauthorized content use.
Meta AI Reaches 1 Billion Monthly Active Users
Category: Business & Industry Trends
🚀 Meta AI has reached 1 billion monthly active users across its apps, doubling from 500 million in September 2024.
🗣️ CEO Mark Zuckerberg emphasized personalization, voice conversations, and entertainment as key focus areas for Meta AI's development this year.
💸 Meta is considering paid recommendations or a subscription service for Meta AI, aiming to compete with other AI chat apps like ChatGPT.
DeepSeek Unveils Compact R1 AI Model That Runs on a Single GPU
Category: Research & Development
🖥️ DeepSeek's distilled R1 AI model (DeepSeek-R1-0528-Qwen3-8B) can run on a single GPU, reducing computational requirements compared to full-sized models.
📊 The model outperforms Google's Gemini 2.5 Flash on the AIME 2025 math benchmark and nearly matches Microsoft's Phi 4 on HMMT math skills tests.
📝 Released under a permissive MIT license, DeepSeek-R1-0528-Qwen3-8B is available for commercial use and accessible via several API hosts.
This content is curated and compiled autonomously using a Python-driven AI system I wrote. While every effort is made to ensure accuracy, I welcome your feedback on any discrepancies you may notice.