Stories worth reading
Meta introduces Muse Wonder for agentic tasks
Meta announced Muse Wonder and Muse Wonder Full, models aimed at complex multi-step tasks ranging from document work to building applications.
The UK strengthens defences against autonomous AI risks
The UK government outlined measures to use advanced AI in cyber defence and prepare for emerging risks from autonomous systems.
Microsoft introduces Maia 300 and a new bet on custom AI infrastructure
Microsoft is expanding its custom AI infrastructure program. The company introduced Maia 300 and outlined plans to scale Azure Maia to multi-gigawatt deployments from 2027.
Google Research@ 2026: AI for science, health and global challenges
On October 14, Google Research presented new directions in AI research, from scientific discovery to tools intended to support health and public-safety work.
Read article →Tropfest clarifies its generative-AI policy for short films
Tropfest published entry rules and a policy for generative AI
Read article →Codex introduces Composer predictions for the next message
OpenAI introduced Composer predictions for the next message in Codex
Read article →Google Research examines privacy and security for autonomous AI agents
Google Research published a report on privacy and security challenges in agentic systems
Read article →ChatGPT camera combines scanned pages into a single PDF
ChatGPT made it easier to scan notes and multipage documents
Read article →ChatGPT adds virtual try-on for clothing from a photo
OpenAI described virtual try-on for clothes and accessories in ChatGPT
Read article →ElevenReader expands voice reading and audiobook access in Brazil
ElevenLabs announced an expansion of ElevenReader in Brazil, bringing local voices and audiobook access to a broader audience.
Read article →Runway, Tech Coalition, and Lantern join efforts to improve generative-media safety
Runway described an initiative with the Tech Coalition and Lantern focused on safer generative media and shared industry practices.
Read article →Anthropic investigates unintended actions by autonomous AI agents
Anthropic examined cases in which an agent model may take an unintended path toward a goal and outlined practical lessons for controlling autonomous systems.
Read article →