⚡Microsoft Unveils BitNet.cpp, Run 100B LLMs on a Single CPU!💡💻

🛡️Meta Fights Fraud with Facial Recognition Tech!🔍📸

In Today's Blog:

• 📸YouTube’s New Label Tackles AI Content!✨
  

• 🚫Meta’s New Facial Recognition to Fight Celeb-Bait Ads!👤

• 🎨Google Gemini’s New AI Image Magic: Resize, Edit, and More!✨

• 🚀Microsoft Unveils BitNet.cpp: Run 100B LLMs on a Single CPU!💻

📸YouTube’s New Label Tackles AI Content!✨

YouTube has introduced a new label, “captured with a camera,” to identify videos that have been verified as using real cameras without any alterations.

• 📹 YouTube Shorts can now be up to 3 minutes long.
• 📸 Videos captured with a real camera will have a “captured with a camera” label to distinguish them from AI-generated content.
• 🏷️ This label requires the use of C2PA-supported tools (version 2.1 or higher) to verify the video’s origin.
• ✂️ The label won’t apply if the video has been significantly edited or if the metadata chain is broken.

      Read about it here: Link

🚫Meta’s New Facial Recognition to Fight Celeb-Bait Ads!👤

Meta is testing facial recognition technology to combat scam ads that misuse celebrity images, known as "celeb-bait." This technology will compare faces in ads against public figures’ profile pictures on Facebook and Instagram to identify and block scams.

• 🕵️ Meta is testing facial recognition to detect and block scam ads using celebrity images.
• 📸 Meta is also trialing video selfies for faster account recovery if users get locked out. 
• 🔒 Meta claims it deletes facial data immediately after use and doesn’t store it for other purposes. 
• 🌍 These tests are not currently being conducted in the U.K. or EU due to stricter data protection laws.

     Read more about it: Link

🎨Google Gemini’s New AI Image Magic: Resize, Edit, and More!✨

Google has updated its AI chatbot, Gemini, with the new Imagen 3 text-to-image model. Additionally, Gemini can now edit specific areas of generated images, similar to Adobe’s Generative Fill feature.

• 📸 Google has updated its AI chatbot, Gemini, with the new Imagen 3 text-to-image model.
• 🖼️ Gemini may soon allow users to generate images in various aspect ratios like 16:9, 4:3, and 3:4. 
• ✏️ New features might enable users to select and edit specific areas of generated images, similar to Adobe’s Generative Fill.
• 🗣️ Google has rolled out Gemini Live, allowing users to engage in conversations with the AI chatbot.

     Read about it here: Link

🚀Microsoft Unveils BitNet.cpp: Run 100B LLMs on a Single CPU!💻

Microsoft has launched BitNet.cpp, an inference framework for 1-bit large language models (LLMs), enabling efficient and fast processing on local devices. This framework can run a 100B BitNet b1.58 model on a single CPU at speeds comparable to human reading.

• 🖥️ Microsoft has launched BitNet.cpp, an inference framework for 1-bit large language models (LLMs).
• 🚀 BitNet.cpp can run a 100B BitNet b1.58 model on a single CPU, achieving speeds of 5-7 tokens per second.
• 🌱 The framework significantly reduces energy consumption by up to 82.2% on x86 CPUs.
• 🔮 Plans include support for NPU and GPU and further development of 1-bit LLMs in large-scale settings.

     Read about it here: Link

Other stuff

• 🌍Scope3 Tackles AI’s Carbon Footprint with $25M Boost!💡 [Link] RECOMMENDED
• 🤖Zoom’s AI Avatars: A Game-Changer or Deepfake Danger?🛡️ [Link] RECOMMENDED
• 🏠OpenAI Expands Globally! New Offices in Singapore, Paris, and More!🌍 [Link] 
• 🚀Atlassian Unleashes Rovo: Your New AI Teammate!✨ [Link] RECOMMENDED

revideo: Create Videos with Code [Link]

Dream Machine: An AI model that makes high-quality, realistic videos fast from text and images. [Link]

☀️☀️DepthSplat: Gaussian Splatting + Depth

ETH unveils DepthSplat the new SOTA in depth estimation and novel view synthesis tasks. The key feature is the cross-task interactions between Gaussian Splatting & depth estimation.