AI Native Foundation Weekly Newsletter: 18 April 2025
Contents
-
OpenAI o3 and o4-mini: Smarter Models with Full Tool Access
-
Meta Enhances EU AI with Public Content Training, Offers Clear Opt-Out
-
NVIDIA Launches First US-Built AI Supercomputers with Advanced Manufacturing
-
ChatGPT Memory: Personalized AI that Learns About You
-
Claude’s Research & Google Workspace Integration Expands AI Capabilities
-
Create 8-Second Videos with Veo 2 in Gemini Advanced & Whisk
-
Kling AI 2.0: Revolutionary Video & Image Editing Breakthroughs
OpenAI o3 and o4-mini: Smarter Models with Full Tool Access
OpenAI’s most intelligent models to date combine advanced reasoning with seamless tool integration. o3 reduces major errors by 20% compared to o1, while o4-mini delivers exceptional performance for its size, achieving 99.5% accuracy on AIME 2025. Both models can leverage web search, code interpretation, and visual analysis. Available now to ChatGPT Plus, Pro and Team users (replacing o1 and o3-mini), with Enterprise and Edu access coming in one week. Free users can access o4-mini via the ‘Think’ option.
Meta Enhances EU AI with Public Content Training, Offers Clear Opt-Out
Meta is training AI models using public content from adult EU users and their interactions with Meta AI to better reflect European cultures and languages. The company is providing transparent notifications and an easy-to-find objection form, respecting European privacy regulations while excluding private messages and minors’ data. This approach aims to create AI that’s not just available to Europeans but specifically built for their diverse communities.
NVIDIA Launches First US-Built AI Supercomputers with Advanced Manufacturing
NVIDIA is partnering with leading manufacturers to produce AI supercomputers entirely in the US for the first time, with Blackwell chip production starting in Arizona and assembly plants opening in Texas. This domestic manufacturing initiative leverages NVIDIA’s own AI technologies, including Omniverse digital twins and Isaac GR00T robotics, to automate production. The company plans to produce up to $500 billion in AI infrastructure within four years, strengthening supply chains and creating hundreds of thousands of jobs for America’s growing AI economy.
ChatGPT Memory: Personalized AI that Learns About You
ChatGPT’s memory feature creates personalized interactions by remembering details between conversations. You can teach it preferences by simply stating them or ask what it remembers. Memory works through saved memories and chat history references, with Plus/Pro accounts having access to both while Free accounts only get saved memories. Memory also enhances web searches by using what it knows about you. You maintain complete control with options to delete specific memories, reset all memory data, or turn the feature off entirely through settings or by using Temporary Chats.
Claude’s Research & Google Workspace Integration Expands AI Capabilities
Claude introduces two powerful features that transform AI collaboration: Research, which conducts multi-step web searches with citations, and Google Workspace integration connecting Gmail, Calendar, and Docs. These tools help marketing teams accelerate product launches, sales teams prepare for client meetings, and engineers create technical solutions. Research is available in beta for paid plans in the US, Japan, and Brazil, while Google Workspace integration is available in beta to all paid users, though team/enterprise admins must enable domain-wide access first. Document cataloging provides enterprise users with more precise information retrieval across organizational knowledge.
Create 8-Second Videos with Veo 2 in Gemini Advanced & Whisk
Transform text prompts into cinematic 720p videos using Google’s powerful Veo 2 technology, now available for Google One AI Premium subscribers. Generate realistic videos with fluid motion in Gemini Advanced or animate your images with Whisk Animate. All videos include SynthID watermarking for transparency. Perfect for creating short social media content, visualizing creative concepts, or bringing your imaginative scenes to life.
Kling AI 2.0: Revolutionary Video & Image Editing Breakthroughs
Kling AI’s Phase 2.0 delivers advanced multimodal editing with upgraded KLING and KOLORS models. The standout Multi-Elements Editor enables seamless adding/swapping/deleting of video content, while new Image Editing and Restyle features give creators unprecedented control. These tools transform storyboarding, retouching, and creative workflows with enhanced prompt adherence and cinematic quality.