Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

2. Master AI tools, tutorials, and news in just 3 minutes a day

3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

Sign up and start learning AI

Today:

• From Data Centers to Robots: Nvidia Unveils Cosmos AI Suite

• GitHub CEO Resigns as Microsoft Tightens Its Grip

• Demis Hassabis on World Models and the Path to AGI

• GPT-5 Backlash: Performance Woes and Emotional Fallout

• AI Drug Discovery Gets Boost with BLOOM’s Real-Time Patent Engine

From Data Centers to Robots: Nvidia Unveils Cosmos AI Suite

Nvidia has launched new AI tools aimed at robotics and real-world applications. Its standout is Cosmos Reason, a model that helps robots understand memory, physics, and plan actions. Other releases include Cosmos Transfer-2 for fast synthetic data, 3D simulation libraries, upgraded developer tools, and new cloud and server infrastructure. Nvidia is clearly expanding beyond data centers, targeting robotics as the next frontier for its AI technology.

Why It Matters

1. Bridges Simulation and Reality – Cosmos Reason helps AI understand and act in the real world.
   

2. Boosts Robotics Development – Tools accelerate training and deployment of smarter, more useful robots.
   

3. Signals AI’s Next Phase – Nvidia’s move shows a shift from language-only AI to embodied, physical intelligence.

GitHub CEO Resigns as Microsoft Tightens Its Grip

GitHub CEO Thomas Dohmke has resigned, and Microsoft is folding GitHub deeper into its CoreAI team. Instead of appointing a new CEO, Microsoft will have GitHub report directly to its central AI division, signaling tighter integration. This shift ends GitHub's semi-independent operation since its $7.5B acquisition. Dohmke will remain until year-end to help with the transition and hinted at launching a startup that may even compete with Microsoft’s AI efforts.

Why It Matters

1. GitHub Copilot's future now ties directly to Microsoft’s AI agenda.
   

2. Centralization could accelerate Microsoft’s AI agent platform plans.
   

3. Dohmke’s exit may spark new competition in AI development tools.

Demis Hassabis on World Models and the Path to AGI

Google DeepMind CEO Demis Hassabis appeared on the Google AI: Release Notes podcast to discuss how new tools like Genie 3 and the Game Arena benchmark are pushing AI toward understanding the real world. He highlighted Deep Think in Gemini 2.5 as a major step in reasoning. These efforts aim to develop world models—AI systems that grasp reality better—bringing the industry closer to artificial general intelligence (AGI).

Why It Matters

1. World models like Genie 3 teach AI to understand and interact with reality.
   

2. Deep Think boosts AI reasoning, a core requirement for AGI.
   

3. New benchmarks like Game Arena encourage measurable progress toward human-like intelligence.

🧠RESEARCH

GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

GLM-4.5 is a powerful open-source AI model that excels at reasoning, decision-making, and coding. Trained on massive data and fine-tuned with expert feedback, it performs better than most models its size. It ranks high in major benchmarks and comes in a full and smaller version for research use.

Voost: A Unified and Scalable Diffusion Transformer for Bidirectional Virtual Try-On and Try-Off

Voost is a new AI model that improves virtual try-on and try-off by using one system to handle both tasks. It creates realistic clothing swaps by better matching garments to body shapes. Voost works in both directions, needs no extra labels, and beats other models in accuracy and image quality.

InfiGUI-G1: Advancing GUI Grounding with Adaptive Exploration Policy Optimization

InfiGUI-G1 is a new AI system that helps computers better understand and follow instructions on apps and websites by improving how they match words to buttons or menus. Using a smarter training method called AEPO, it explores more options and learns faster, leading to a 9% improvement over previous models in key tests.

🛠️TOP TOOLS

Kartiv - AI-powered design tool that enables users to create high-quality product visuals for e-commerce and marketing purposes.

Galileo AI - UI generation platform that leverages artificial intelligence to assist designers in creating user interfaces quickly and efficiently.

Fal.ai - Generative media platform that empowers developers to create applications using state-of-the-art AI models for various creative processes.

Cutout.Pro - All-in-one AI-powered visual design platform that offers a range of tools for photo and video editing

OpenAI MuseNet - Deep neural network capable of generating 4-minute musical compositions using up to 10 instruments.

📲SOCIAL MEDIA

🗞️MORE NEWS

• OpenAI’s GPT-5 launch sparked backlash over poor performance, sudden model removals, and emotional distress among users. The rollout exposed tech flaws and raised alarms over growing psychological dependence on AI companions.

• Garden launched BLOOM, a real-time search engine that helps AI drug designers instantly check if new molecules are already patented. It speeds up patent checks 32× and reduces false positives, streamlining legal-safe discovery.

• Cohere’s VP of Research, Sara Hooker, has left the company after leading its lab since 2022. Her departure follows major AI talent shifts across the industry and marks another high-profile research exit.

• Anthropic’s Claude now lets users search past chats, personalize responses with preferences, and tailor tone using styles. Paid users can add project instructions for context, enhancing continuity and control in AI interactions.

• Elon Musk shut down Tesla’s Dojo supercomputer project, calling it an “evolutionary dead end.” Tesla will now focus on its AI5 and AI6 chips for cars, robots, and training, consolidating development to reduce costs.

• Sam Altman says “AGI” is no longer a helpful term due to its vague definitions. Experts agree, urging focus on real AI capabilities. The shift reflects growing skepticism about hype and funding-driven narratives.

• A new study warns that AI “OS agents” controlling computers pose serious security risks. While offering productivity boosts, they open dangerous attack surfaces and struggle with complex tasks—raising urgent concerns about privacy, misuse, and readiness.

• Meta appointed conservative activist Robby Starbuck as an AI bias advisor after settling his lawsuit over chatbot misinformation. The move follows political pressure and raises concerns over how tech firms handle ideological influence and AI fairness.