This weekend in Generative Media

Anthropic to pay $1.5B settlement; AI in museums; Welles estate slams AI company

News and Opinion

• Anthropic to pay authors $1.5 billion to settle lawsuit over pirated books used to train AI chatbots (AP)

  • 5 Lessons Of Anthropic's Massive $1.5 Billion Settlement: The Largest In US Copyright History (Peter Csathy on Substack)

  • Screw the money — Anthropic’s $1.5B copyright settlement sucks for writers (TechCrunch)

• A.I. Enters the Museum (New York Times)

• Orson Welles Estate Slams AI Company’s Plan to Recreate Lost Scenes of ‘Magnificent Ambersons’: ‘Disappointing’ and an ‘Attempt to Generate Publicity’ (Variety)

• Chinese social media platforms roll out labels for AI-generated material (Engadget)

• Don’t look now, but there’s an AI-generated Italian teacup on your child’s phone. What does it mean? (AP)

• The Doctors Are Real, but the Sales Pitches Are Frauds (New York Times)

• Substack CEO says AI can drive a slop future or a cultural renaissance (Business Insider)

• When It Comes to Spotting Fake Receipts, It’s A.I. vs. A.I. (New York Times)

Software, Hardware and Datasets

• We’re excited to share that Bria’s enterprise-grade Visual GenAI 3.2 Text-to-Image model is now natively integrated with 🤗 Diffusers 🧨, no custom pipeline required. (LinkedIn)

• Nano Banana is NOW LIVE alongside FLUX Premium in LTX Studio. (X)

• New from Meshy: 3D to Video. (X)

• A new Qwen Image LoRA "Boring Reality" was just released on Huggingface by author kudzueye (X)

• These AI glasses hit $1M in pre-orders in 72 hours and have features missing from Meta Ray-Bans (ZDNet)

• Introducing EditNet — a high-quality before/after dataset we built to train our models, named in homage to ImageNet. (project page)

Research

• LuxDiT: Lighting Estimation with Video Diffusion Transformer (project page)

• Complete Gaussian Splats from a Single Image with Denoising Diffusion Models (project page)

• GRMM: Real-Time High-Fidelity Gaussian Morphable Head Model with Learned Residuals (project page)

• FastAvatar: Instant 3D Gaussian Splatting for Faces from Single Unconstrained Poses (arXiv)

Misc

• Nano Banana Hackathon: Compete for your share of over $400,000 in prizes (Kaggle)

• This tutorial explains how to grab Motion-Captures from your Webcam or any video file, using the free BlendARMocap add-on for Blender by cgtinker with Google Mediapipe, and turn them into 3D-animated Controlnets, using the OpenPoseBone tool by toyxyz. (YouTube)

• I’ve decided this is how I learn now (Threads) [Sound on!]