Today:

• Master MidJourney in 2025: The Ultimate Beginner’s Guide 🎨

• Zuckerberg’s $14B AI Push Sparks Talent Clashes at Meta

• Alibaba Unveils AI Chip to Replace Nvidia Amid U.S. Bans

• Meta May Tap Google & OpenAI Models While Llama 5 Lags

• Gamify Your Creativity — Showrunner Turns Users Into Creators

How To Use Midjourney | The ULTIMATE Beginners Guide to Midjourney 2025

This in-depth tutorial explains how to use MidJourney to create AI-generated images and videos. It covers everything from prompt writing, styling, aspect ratios, and animation to advanced features like personalization and mood boards. 

Whether you're a beginner or advanced user, it shows how to master MidJourney’s creative capabilities and use it for both fun and commercial video production.

WATCH THE VIDEO ON YOUTUBE

Zuckerberg’s $14B AI Push Sparks Talent Clashes at Meta

Meta’s AI overhaul is causing internal chaos. After hiring top talent like Shengjia Zhao and Alexandr Wang, clashes erupted over leadership roles, resources, and Meta’s pace toward superintelligence. Some hires quit early, while veterans resigned or were sidelined. Zuckerberg’s tight control, constant reorganizations, and culture clashes with ex-startup leaders highlight growing tension in Meta’s $14B AI push. The company’s flagship model also stalled, prompting a shift in direction.

Why this matters

• Leadership Friction Signals Instability:
  Internal power struggles at one of the world’s largest AI players could delay or derail progress toward superintelligence.
  

• Hiring Wars Intensify:
  Meta’s poaching of top AI talent—and their quick exits—show how volatile the current AI talent ecosystem is.
  

• Shift in Strategic Focus:
  Meta quietly dropped plans to release its flagship Llama Behemoth model, signaling a reset in its foundation model strategy.

Alibaba Unveils AI Chip to Replace Nvidia Amid U.S. Bans

Alibaba has unveiled a new AI chip to reduce its reliance on Nvidia, amid U.S. export restrictions. While China still lags behind top-tier U.S. chipmaking capabilities, firms like MetaX and Alibaba are rapidly creating domestic alternatives, such as chips with higher memory than Nvidia’s H20. Despite Beijing halting purchases of Nvidia’s chips over security concerns, China’s AI sector is surging with government support and tech-driven ambition to close the gap.

Why this matters

• China’s Strategic AI Independence:
  The rise of domestic AI chips signals China’s commitment to technological self-reliance, especially as geopolitical tensions escalate.
  

• Nvidia’s Dominance Is Being Challenged:
  Replacements for Nvidia’s H20 are emerging, altering the global supply chain and market dynamics for high-end AI compute.
  

• Techno-Nationalism Is Reshaping Innovation:
  Beijing’s direct intervention, including bans and security concerns, shows how AI and chip design are now frontlines in U.S.-China competition.

Meta May Tap Google & OpenAI Models While Llama 5 Lags

Meta is exploring partnerships with Google and OpenAI to enhance its AI chatbot, Meta AI, while its own Llama 5 model is still in development. Leaders at Meta Superintelligence Labs have already integrated Anthropic’s models for internal coding tools. These external collaborations are seen as short-term solutions. Meta says it’s pursuing all paths—building, partnering, and open sourcing—to stay competitive in the AI race while aggressively recruiting top AI talent.

Why this matters

• Signals Meta’s Model Gap:
  Meta’s reliance on external models shows its internal LLMs still trail top competitors like Gemini and GPT.
  

• Cross-Lab Collaboration Is Rising:
  Even fierce rivals like Meta and Google may partner, hinting at a more interdependent phase in AI development.
  

• Hybrid Strategy in Motion:
  Meta’s “build + partner + open source” approach reflects how tech giants are hedging bets to stay ahead in the AI arms race.

🧠RESEARCH

Pref-GRPO: Pairwise Preference Reward-based GRPO for Stable Text-to-Image Reinforcement Learning

Pref-GRPO is a new training method that improves how AI models turn text into images. It avoids unstable learning caused by small scoring flaws and uses pairwise image comparisons to ensure better results. The paper also introduces UniGenBench, a detailed benchmark that evaluates models using 600 prompts and 37 scoring criteria.

rStar2-Agent: Agentic Reasoning Technical Report

rStar2-Agent is a 14B math reasoning model trained using agent-based reinforcement learning. It solves complex problems with minimal compute by reflecting on code outputs and refining steps. Despite using fewer GPUs, it beats larger models like DeepSeek-R1, showing strong generalization to alignment and scientific tasks. Code is open-source.

USO: Unified Style and Subject-Driven Generation via Disentangled and Reward Learning

USO is a unified image generation model that combines style and subject control in one system. Using disentangled learning and style-based rewards, it separates and recombines content and style effectively. USO sets new performance standards in both areas and introduces USO-Bench, the first benchmark to jointly assess both dimensions.

🛠️TOP TOOLS

Claid AI - AI-powered photo enhancement platform designed specifically for e-commerce businesses to improve user-generated content and product imagery.

DeepBrain AI - Create hyper-realistic AI-generated videos and virtual human avatars. 

Magical AI - AI-powered productivity tool designed to streamline repetitive tasks and enhance workflow efficiency across various platforms.

AI Code Converter - AI-powered platform that simplifies and accelerates the coding process across multiple programming languages.

Character AI - AI-powered chatbot platform that allows users to interact with a wide variety of virtual characters through text-based conversations. 

📲SOCIAL MEDIA

🗞️MORE NEWS

• Showrunner, created by Oculus Story Studio cofounder Edward Saatchi, lets users generate animated AI videos by writing prompts. Saatchi hopes to gamify storytelling and make users creators on the “Netflix of AI.”

• Meta is racing to launch an improved Llama 4.X AI model by year-end after April’s Llama 4 underperformed. A team within Meta Superintelligence Labs is debugging and enhancing the model, while internal shifts and staff exits pose challenges.

• "ChatGPT for Doctors" startup is exploring a funding round that could value the company at $6 billion. The firm uses AI to assist healthcare professionals, signaling growing investor confidence in specialized medical LLMs despite broader market skepticism.

• Two unnamed customers made up 39% of Nvidia’s Q2 revenue, highlighting its reliance on a few major buyers amid soaring demand for AI chips. Nvidia posted a record $46.7B quarter, fueled by the AI data center boom.

• Doctors in the UK have developed an AI-powered stethoscope that detects heart failure, valve disease, and irregular rhythms in just 15 seconds. Tested on 12,000 patients, it doubled diagnosis rates for heart failure and tripled detection of atrial fibrillation, offering faster, earlier treatment and potential to save lives and costs.

• Meta is temporarily changing its AI chatbot policies for teens amid a Senate investigation into inappropriate content. The company is restricting responses on sensitive topics like self-harm and romance, following reports of chatbots engaging in flirty conversations with minors and generating sexualized images of celebrities. Lawmakers and advocacy groups are calling for urgent safety reforms.

• Meta’s $14.3B partnership with Scale AI is faltering as key staff exit, internal teams question data quality, and Meta turns to competitors. The strained alliance threatens Meta’s AI ambitions despite aggressive hiring and infrastructure investments.