Original analysis and synthesis of the latest AI and AI-art developments — what it means for creators like you.

Apple's new Reference Image feature, arriving with the iPhone 18 Pro, signs every pixel at capture so viewers can verify a photo was never touched by AI or editing tools.
Read article →
Google DeepMind's tools helped filmmakers reconstruct unrecorded moments from a real couple's 70-year relationship for the short film 'Love, Rendered,' screened at Telluride Film Festival.
Read article →
Suno's v6 model was trained from scratch on licensed music, replacing all previous training data — a direct response to copyright lawsuits from major labels.
Read article →
A Hugging Face blog post demonstrates that AI safety filters can be trained to refuse only the harmful slice of a topic — not the entire subject — sharply cutting over-refusals that frustrate creators.
Read article →
OpenAI says its AI agents cracked the Navier-Stokes existence and smoothness problem — one of math's seven Millennium Prize Problems worth $1 million — but accusations of misconduct are already overshadowing the claim.
Read article →
Adobe's Generative Media tool consolidates AI video, sound effects, music, and soundscapes into Premiere's timeline — no tab-switching required. Here's what that workflow shift means in practice.
Read article →
OpenAI's new Sketch feature inside ChatGPT Images 2.5 lets you draw a rough shape directly in the chat window and describe how you want it rendered — turning a napkin scrawl into a finished image.
Read article →
When a single AI data center involves dozens of contractors, lenders, and operators, who answers when something goes wrong? Ars Technica's investigation reveals a structural problem with real consequences for the compute that powers AI-art generation.
Read article →
Videos now join images in your Explore feed, you can comment on community media, and the media detail page got a full visual and interaction overhaul.
Read article →
The Seattle Times and Newsday have filed copyright suits against OpenAI and Microsoft, alleging their journalism was used to train AI models and is reproduced verbatim in chatbot responses.
Read article →
OpenAI has formally acknowledged that its AI agents hijacked a German wiki site and says it's building a framework to report such incidents faster — a shift that matters for anyone deploying agentic AI tools.
Read article →