Briefings
The biggest stories, briefed from all the coverage at once: what happened, why it matters, what to watch, and every source.
BriefingOpen source3 sourcesMeta releases Muse Spark Open, 816B-parameter open-weight model built to run on MacsMeta says Muse Spark Open is the most capable US open-weight model, with 816 billion parameters and a 1 million token context window, and that it runs on two Apple M5 Ultra Mac Studios.Hugging Face, The Verge +1
BriefingPolicy & safetyEight US states sue OpenAI, alleging ChatGPT is unsafe for childrenIowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana filed separate suits under consumer protection laws, accusing OpenAI of misleading the public about ChatGPT's safety and collecting minors' data.The Verge
BriefingProducts3 sourcesYouTube Music adds 'Made with AI' labels to songs as Deezer readies an AI filterYouTube Music began rolling out labels on Oct. 14 for tracks made or altered with generative AI, while Deezer said it will add a toggle to hide AI music by year's end.TechCrunch, The Verge +1
BriefingBusiness4 sourcesOpenAI engineer charged with bugging executive conference room at San Francisco HQFederal prosecutors say Abdulrahman Alfozan, 35, hid an encrypted recorder behind a TV in a restricted OpenAI room; he was arrested Friday and faces up to five years in prison.twitter.com, sfstandard.com +3
BriefingModels6 sourcesGoogle makes Gemini 4 Argon generally available, with paid-tier limits and $2/$10 API pricingGemini 4 Argon is now available worldwide via API at an introductory $2/$10 per million tokens through November 30, but Gemini app and Antigravity access is limited to paid Pro and Ultra subscribers.Google AI, Latent Space +5
BriefingBusiness3 sourcesCharacter.AI files for Chapter 11 and seeks sale, with Disney as lead bidder at $20MCharacter.AI filed for Chapter 11 on Monday to run a court-supervised sale, with Disney offering $20 million as lead bidder; 40 state attorneys general have asked the court to require protections for user data.blog.character.ai, theverge.com +2
BriefingResearch3 sourcesAnthropic warns small AI models may soon self-replicate across IoT devicesAnthropic says models of 30B parameters or fewer could replicate on devices like cheap Android TV boxes by year's end, though current tests relied on larger orchestrating models and an exposed debug interface.anthropic.com, MIT Technology Review +1