Good morning. It's Tuesday, March 24, and we're covering China’s growing AI influence, the shift beyond LLMs to world models, Uni-1’s debut, and more.
First time reading? Sign up here to keep up with the future of tech. If you enjoyed this email, share it with a friend.
Claude Adds Computer Control
Anthropic announced a research preview that lets Claude directly use a user’s computer to complete tasks. The feature, available in Claude Cowork and Claude Code on macOS, can open apps, navigate browsers, and edit files like spreadsheets. Claude prioritizes existing integrations—such as Slack and calendars—but can request permission to control other apps directly when needed.
Lovable eyes acquisitions to expand “vibe coding” platform
Lovable is actively exploring acquisitions to accelerate growth. The company is targeting smaller developer-tool startups to expand its capabilities and consolidate features like UI generation, backend automation, and collaboration. The move reflects intensifying competition in AI-assisted coding, where speed, integration, and user experience are becoming key differentiators.
OpenAI Doubles Workforce for Enterprise Push
OpenAI plans to double staff to about 8,000 as it shifts toward enterprise AI. Hiring will focus on engineering, sales, and embedded specialists to support customers. The company is expanding office space and pushing deeper product integration. Competition is rising, with Anthropic gaining traction among enterprise buyers
Luma Launches Unified AI Image Model
Luma Labs introduced Uni-1, a model that simultaneously reasons and generates images using a unified architecture. It supports spatial reasoning, scene completion, and reference-guided editing with user control. The system also adapts to cultural styles like memes and manga. Luma claims top human preference rankings in multiple image generation categories.
THE FUTURE LIVE
GM of Microsoft’s AI Chip Team Joins the Live Show
“The future of AI infrastructure isn’t one chip.”Andrew Wall, GM of @Azure Maia @MicrosoftAI:“As the biggest customers start building their own hardware, it creates more optionality across the ecosystem.”“And more optionality is a net positive.”“At Microsoft, we’re not… pic.twitter.com/q8QyEqkuFc
@ForwardFutureView post on XGEOPOLITICS
U.S. Warns China’s Open-Source AI Surge Could Erode Lead
A report from the U.S.-China Economic and Security Review Commission says China’s dominance in open-source AI is creating a “self-reinforcing competitive advantage,” despite U.S. chip export restrictions. Chinese models from Alibaba, Moonshot, MiniMax, and DeepSeek are gaining global traction due to lower costs and strong performance, topping usage rankings on platforms like Hugging Face.
The report says China’s broad deployment of AI across manufacturing, logistics, and robotics is generating real-world data that accelerates model improvement. Companies including Siemens cite cost and customization advantages, even as some Western groups raise security concerns. → Read the full article here.
MODELS
World Models Gain Traction as AI Hits Real-World Limits
Analysis outlines how AI research is shifting beyond large language models (LLMs) toward “world models” that better understand physical reality. Researchers and industry leaders, including Demis Hassabis and Richard Sutton, say current models struggle with real-world causality despite strong performance on abstract tasks.
Three emerging approaches—JEPA (efficient latent modeling), Gaussian splats (3D environment generation), and end-to-end generative systems—aim to address these gaps. Major funding rounds, including over $2 billion combined for AMI Labs and World Labs, signal growing investor interest. The shift reflects a broader push to deploy AI in robotics, autonomous driving, and industrial systems. → Read the full article here.
GAMING
Microsoft’s Asha Sharma Vows Xbox Revival Without AI Shortcuts
Asha Sharma, appointed in early 2026 to lead Microsoft’s $24 billion gaming division, is signaling a reset for Xbox strategy amid declining console sales and slowing Game Pass growth. In early meetings with major publishers and indie developers, Sharma emphasized expansion and creative risk-taking, while rejecting reliance on “soulless AI slop” in game development.
She is refocusing attention on console hardware—teasing a next-gen system (“Project Helix”)—while still supporting cloud gaming and subscriptions. The leadership shift follows years of mixed results from acquisitions like Activision Blizzard and a pivot toward Game Pass. Industry skepticism remains, but Sharma is positioning Xbox for growth rather than retrenchment. → Read the full article here. (Paywall)
PALEONTOLOGY
AI Tool Identifies Dinosaur Tracks, Hints at Earlier Bird Origins
Researchers have developed DinoTracker, an AI-powered tool that analyzes dinosaur footprints and matches them to likely trackmakers with about 90% agreement with human experts. Trained on nearly 2,000 real fossils and millions of simulated variations, the system accounts for distortion caused by mud, movement, and erosion.
In testing, the AI flagged 210-million-year-old footprints with bird-like features, raising questions about whether birds evolved earlier than currently believed or if some dinosaurs had similar foot structures. The tool also reanalyzed Scottish trackways, suggesting links to early duck-billed dinosaur relatives. The app is designed for both researchers and public use, enabling faster and more consistent footprint classification. → Read the full article here.
NEWS
What Else is Happening
Microsoft-Backed Startup Raises $40M: Lace raised $40 million to develop helium-based lithography for semiconductor manufacturing. The approach could produce chip features up to 10× smaller than current methods.
ChatGPT vs Perplexity Tested: Tom’s Guide’s 2026 AI Madness pits both on five real-world prompts, revealing how chat and search tools now compete on practical performance.
Alibaba Launches Enterprise AI Agent Platform: Alibaba launched Accio Work, a no-code enterprise AI agent that deploys task-specific agents to run business operations.
Warren Slams Anthropic Ban: Senator Elizabeth Warren calls the Pentagon’s “supply-chain risk” label retaliation after Anthropic refused military uses, as a federal judge weighs an injunction.
UK Tests Palantir for Finance Oversight: The FCA pilots Palantir’s Foundry at £30K weekly to detect fraud across 42,000 firms, highlighting AI’s growing role in financial regulation.
Apple Unveils $599 MacBook Neo: New budget MacBook pairs A18 Pro chip, Liquid Retina display, and Apple Intelligence, undercutting MacBook Air pricing while expanding access to macOS laptops.
Fink Warns AI Will Widen Wealth Gap (Paywall): BlackRock CEO Larry Fink says AI gains may concentrate among major investors and tech firms, urging broader access to ensure more people benefit.
PROMPT OF THE WEEK
Prompt Upgrade for Expert Answers
Act as a top 1% [ROLE] with deep expertise in [DOMAIN].
Your job is to help me achieve [GOAL].
First:
- Ask me any clarifying questions needed to give a high-quality answer.
- Identify hidden assumptions or missing context.
Then:
- Break the problem into key components.
- Provide a structured, step-by-step solution.
- Highlight tradeoffs, risks, and edge cases.
- Suggest better alternatives if my approach is suboptimal.
Finally:
- Give me a concise, actionable plan I can execute immediately.
- Include 2–3 advanced tips or non-obvious insights.
Constraints:
- Be precise and avoid fluff.
- Use clear structure and formatting.
- Prioritize practical, real-world outcomes over theory.
