Weekly AI #17

Welcome to Weekly AI #17
Last week, Google AI Studio rolls out updates to Gemini 2.0 Flash, enhancing its speed and efficiency for real-time applications, and introduces Gemma 3, their latest open-source model driving accessible AI innovation.

Top 5 Tools

Duck.ai - Anonymous access to popular AI models

EnhanceAI - EnhanceAI empowers creators with AI-driven tools for image generation, upscaling & digital content

Digits AI Accounting - Experience the first end-to-end accounting platform of the AI era

Blotato - Create viral posts and faceless videos

Unreal Speech - Fast & affordable text-to-speech API

Top News / Shorts

Mon, Mar 10
- Pika Labs offers virtual dress code trials for styling
- ElevenLabs reduces Scribe Speech to Text costs by 45%
- Lovable to introduce custom domains for enhanced user experience
- Google AI adds code execution in Gemini API for Python-based reasoning
- Microsoft to updates Copilot with free voice, reasoning, and gaming features
- OpenAI highlights risks of optimizing chain-of-thought in reasoning models
- Luma AI adds favorites feature in DreamMachine
- Hunyuan launches TurboS, a hybrid Transformer-Mamba model for AI efficiency
- Hunyuan launches HunyuanVideoGP for local video generation with minimal VRAM

Tue, Mar 11
- Sam Altman's AI creative writing model sparks debate on art authenticity
- Meta AI's DINOv2 model aids cancer therapy research with organoid imaging
- Luma AI introduces Inductive Moment Matching for efficient diffusion models
- OpenAI launches developer tools to simplify AI agent creation
- Perplexity launches Windows app with voice dictation and keyboard shortcuts
- Alibaba updates Qwen Chat with video understanding and guest mode access
- Fal AI partners with Easel AI to integrate their models
- Harvey launches AI agents for legal workflows, rivaling human lawyers
- Move AI launches Gen 2 spatial motion for advanced 3D motion capture
- Reka AI open sources Flash 3, a competitive 21B parameter reasoning model
- Cartesia AI raises $64M Series A to advance real-time voice AI technology
- NEX introduces Marko, an AI agent for automated content marketing
- TypeScript to rewrite in Go for 10x faster performance by 2025
- X launches Talk To X for real-time database insights and answers
- Anthropic CEO predicts AI could write 90% of code in 3-6 months
- Sudowrite launches Muse, an AI model tailored for fiction writing
- Meta tests its first in-house AI training chip with TSMC
- Hunyuan updates I2V training code and LoRAs for new video model
- Replicate reduces Wan2.1 model prices by 40-50% for faster video generation

Wed, Mar 12
- Google Gemma 3 models now available for users on LM Studio
- Google Gemini 2.0 enables seamless multimodal image editing
- Captions launches Mirage, an AI model for audio-to-video generation
- Lovable 2.0 is on the way with exciting new updates
- Google rolls out experimental image generation in Gemini 2.0 Flash
- Perplexity launches MCP for real-time web search in AI assistants
- Google DeepMind unveils Gemini Robotics models for advanced robot capabilities
- Google DeepMind launches Gemma 3, efficient open-source AI models
- Freepik introduces image-to-video on Google Veo 2 in AI Suite
- Channel42 integrates Google Veo 2 for impressive image-to-video results
- Google holds a 14% stake in Anthropic with a $3 billion investment
- Fal integrates Google DeepMind's Veo 2 for image-to-video generation

Thu, Mar 13
- Suno introduces song covers on iOS for creative style transformations
- Freepik adds Topaz AI Video Upscaler & Enhancer to AI Suite
- Google DeepMind upgrades Gemini with free Deep Research and Flash Thinking
- Anthropic updates API to enhance Claude 3.7 Sonnet efficiency
- ElevenLabs achieves HIPAA compliance for healthcare voice AI
- Lovable introduces custom domains for personalized app branding
- Anthropic researches hidden objectives in language models through audits
- Google Gemini enhances performance and adds personalized features
- Pika Labs introduces Pikaffects for character morphing and transitions
- Microsoft introduces Copilot for Gaming with real-time coaching features
- Bolt.new integrates Figma for seamless app development
- Convergence launches DeepWork, its most advanced AI agent
- OpenAI pushes for AI copyright reform and criticizes Chinese AI models
- OpenAI expands Operator access to Pro users in Europe
- Alibaba introduces VACE for all-in-one video creation and editing

Fri, Mar 14
- Apple admits Siri delays but vows to improve virtual assistant
- Alibaba Qwen sets up GitHub repo for QwQ model evaluation
- Perplexity will announce something on 3-17-25
- DeepSeek restricts R&D employee travel amid heightened scrutiny;
- Hailuo AI launches Image-01 for high-fidelity image and video creation
- OpenAI redesigns Chat Playground as Prompts Playground for better testing

Sat, Mar 15
- Figure launches BotQ factory for mass production of humanoid robots

Sun, Mar 16
- Baidu launches ERNIE 4.5 and X1, intensifying global AI competition