Today in Generative Media

Wikipedia ends AI slop; Deepfakes distort reality; Meta's AI memorized verbatim

News and Opinion

• Wikipedia Won't Add AI-Generated Slop After Editors Yelled At Them (Kotaku)

• Deepfake Scams Are Distorting Reality Itself (Wired)

• Meta's AI memorised books verbatim – that could cost it billions (New Scientist)

• AI Has Already Run Us Over the Cliff (Nautilus)

• This A.I. Company Wants to Take Your Job (New York Times)

• ‘AI fatigue’ is settling in as companies’ proofs of concept increasingly fail. Here’s how to prevent it (Fortune)

• They Asked an A.I. Chatbot Questions. The Answers Sent Them Spiraling (New York Times)

Software and Hardware

• Meta launches a surreal AI video editor (Engadget) Meta AI gains video editing capabilities (TechCrunch)

• HP reveals $24,999 hardware created just for Google Beam (The Verge)

• Genspark AI Browser: Lightning Fast, Ad-Blocking, Full Agentic, Autopilot Mode, MCP Store, and Super Agent That Does Everything for You! (genspark.ai)

• 💗Likable is a fully agentic AI that builds, tests, and previews your

  gradio app as you describe it. (X)

• vui: Small Conversational speech models that can run on device (GitHub)

• Highly Compressed Tokenizer Can Generate Without Training (GitHub)

• A custom node for ComfyUI providing an interactive 3D viewer for 8i volumetric videos (MPD format), allowing frame capture and integration into workflows. (GitHub)

• GauSTAR enables photo-realistic rendering, surface reconstruction, and 3D tracking for dynamic scenes from multiview inputs while handling topology changes (GitHub)

• Dolphin: Document Image Parsing via Heterogeneous Anchor Prompting (HuggingFace)

Research

• Seaweed APT2 (project page)

• Unbounded: A Generative Infinite Game of Character Life Simulation (project page)

• Speedy Deformable 3D Gaussian Splatting: Fast Rendering and Compression of Dynamic Scenes (project page)

• DGS-LRM: Real-Time Deformable 3D Gaussian Reconstruction From Monocular Videos (project page)

• QUEEN: QUantized Efficient ENcoding of Dynamic Gaussians for Streaming Free-viewpoint Videos (project page)

• CrossGen: Learning and Generating Cross Fields for Quad Meshing (project page)

• InstaInpaint: Instant 3D-Scene Inpainting with Masked Large Reconstruction Model (project page)

• CAP4D: Creating Animatable 4D Portrait Avatars with Morphable Multi-View Diffusion Models (project page)

• ProSplat: Improved Feed-Forward 3D Gaussian Splatting for Wide-Baseline Sparse Views (arXiv)

• A Stable Whitening Optimizer for Efficient Neural Network Training (arXiv)

• StreamSplat: Towards Online Dynamic 3D Reconstruction from Uncalibrated Video Streams (arXiv)

Misc

• Kalshi airs fully AI-generated ad on primetime (YouTube)