Gemini 2.0 Flash It's Insane!

Google's latest AI model, Gemini 2.0 Flash, has sparked controversy for its ability to remove watermarks from images, raising concerns about potential copyright violations. While its seamless editing features impress, critics argue that the lack of safeguards could enable misuse. Meanwhile, Nvidia has introduced Groot N1, a cutting-edge AI model for humanoid robotics designed to enhance perception, planning, and movement. With its dual-system cognitive architecture, it marks a major leap toward intelligent, adaptable machines. On the AI reasoning front, LG AI Research has unveiled EXAONE Deep, a compact yet powerful model excelling in mathematics, science, and coding. Open-sourced for developers, it challenges industry giants while prioritizing efficiency and accessibility. Exciting breakthroughs are reshaping AI—let’s dive in!

Google’s Gemini AI Sparks Copyright Controversy!

Google’s new Gemini 2.0 Flash AI model has sparked controversy as social media users discovered its ability to remove watermarks from images, including those from stock media providers like Getty Images. This feature, part of the model's experimental image generation tool available through developer platforms like AI Studio, allows users to edit and generate images seamlessly. However, it lacks sufficient safeguards, enabling tasks like watermark removal and filling in gaps left by the process. While other AI tools offer similar capabilities, Gemini 2.0 Flash stands out for its effectiveness and accessibility. Critics argue this functionality could lead to copyright violations, as removing watermarks without consent is illegal under U.S. copyright law. Competing models like OpenAI’s GPT-4 and Anthropic’s Claude explicitly block such actions, highlighting the need for stricter guardrails in AI tools. Google has acknowledged the issue, stating that using its tools for copyright infringement violates its terms of service and that it is monitoring feedback closely.

Nvidia Unleashes Groot N1: The Future of Humanoid Robotics!

Nvidia has unveiled Groot N1, an open-source AI foundation model for humanoid robotics, at GTC 2025. Building on last year’s Project Groot, Groot N1 is a generalist model trained on synthetic and real data, featuring a dual-system architecture inspired by human cognition. Its "slow-thinking" system enables robots to perceive, reason, and plan actions, while the "fast-thinking" system translates plans into precise robotic movements, including multi-step object manipulation. Designed for adaptability across various humanoid robot forms, Groot N1 is accompanied by simulation frameworks and synthetic data blueprints to aid developers. Nvidia CEO Jensen Huang declared this innovation marks "the age of generalist robotics," as the model aims to advance humanoid robots toward practical, mass-produced applications.

LG’s Game-Changing AI Redefining Reasoning & Efficiency

LG AI Research has launched EXAONE Deep, a state-of-the-art reasoning AI model series that sets new benchmarks in mathematical, scientific, and coding tasks while operating with exceptional efficiency. Available in 32B, 7.8B, and 2.4B parameter variants, the models leverage specialized training techniques like SFT and DPO to outperform competitors—despite their compact size. The flagship 32B model rivals giants like GPT-4o and DeepSeek-R1 (671B) in math benchmarks such as CSAT 2025 (94.5) and AIME 2024 (90.0), using just 5% of the computational resources of larger alternatives. Smaller variants dominate lightweight categories: the 7.8B model scored 94.8 in MATH-500 and 59.6 in AIME 2025, while the 2.4B model leads in on-device AI performance. EXAONE Deep also excels in PhD-level scientific reasoning (66.1 on GPQA Diamond) and coding (59.5 on LiveCodeBench), positioning it as South Korea’s first globally competitive reasoning AI. Released as open-source on Hugging Face, the models aim to accelerate the development of "Agentic AI"—systems capable of autonomous hypothesis formulation and decision-making. With efficiency enabling deployment on consumer-grade GPUs and edge devices, EXAONE Deep marks a milestone in democratizing advanced AI capabilities.

Hand Picked Video

In this video, we'll look at China just dropped QwQ-32B, and it’s making waves in the AI world! This model is being called the smallest yet most powerful reasoning model out there.

Top AI Products from this week

• Intelswift AI Agents Marketplace - Intelswift makes customer support effortless. AI chatbots, Live Chat, and AI Copilot work together to automate responses, assist agents, and provide real-time insights—improving efficiency without losing the human touch.

• Notebooks - Your AI Whiteboard - Your AI Whiteboard to create better content faster⚡ ✨ Add PDFs/YouTube/websites → AI writes like your sources 🎯 Create content that actually sounds like you, not generic AI 🤖 ChatGPT + Claude + DeepSeek in one place

• Bunch for Android - BUNCH is an leadership coach that delivers daily bite-sized tips from world-class experts and leaders to help you boost your impact with your team. Everyday is personalized just for you and takes as little as 2 minutes a day!

• Epiphany - Epiphany is the the fastest, most frictionless way to capture your ideas with voice and create actions with them in tools like Notion, Asana, Todoist, Clickup, Todoist, Obsidian, and more. Stop losing ideas to distraction, and start putting them to work.

• Multi-Agent Builder - Credal enables enterprises to deploy secure, collaborative AI Agents to handle multi-step workflows across an organization’s tools, data and subject matter expertise. Agents can autonomously call other agents and take action while respecting data permissions.

• Elevenreader by ElevenLabs - Read text aloud with ElevenReader app. Listen to free audiobooks and read aloud PDFs, eBooks, and Kindle with the highest-quality voice AI.

This week in AI

• Chirp 3 - Google adds its voice model, Chirp 3, to Vertex AI next week. It features speech-to-text & HD text-to-speech for voice assistants, audiobooks, & more, but with usage restrictions.

• MaTVLM - HUST presents MaTVLM, a hybrid Mamba-Transformer VLM, offering up to 3.6x faster inference and 27.5% less GPU memory usage than the teacher model. Code available on GitHub.

• Stable Virtual Camera - Stability AI releases Stable Virtual Camera, turning 2D images into 3D scenes with realistic depth. The research preview is available on Hugging Face for noncommercial use.

• TxGemma - A collection of Gemma-based open models, aims to enhance AI-powered drug discovery. It understands text & therapeutic entities to predict safety/effectiveness of new therapies. Available this month.