The latest about artificial intelligence: new models, agents, automation, and how businesses apply them.
AI safety takes center stage: Anthropic, OpenAI, and Google pursue standards as Congress demands regulation and risks mount.
Alibaba's Qwen team released Qwen-Image-2.1, a 7-billion-parameter open-weight model for image generation and editing that runs on consumer GPUs.
Four consumers sued Anthropic, OpenAI, SpaceXAI and Google in San Francisco, arguing that coordinated AI pacing is a horizontal agreement to restrain product improvement.
Apple's new CEO described the perfect AI device, then said it already exists as the iPhone, while the J490 home hub waits for release.
Trump promised an AI Force and an AI czar, but a New York Times/Siena poll found 47% of Republicans oppose building data centres for AI.
Marshall Choy of Rebellions says falling inference costs will expand the AI market rather than shrink it, as enterprises stop rationing tokens and API calls.
Dario Amodei's plan to pace frontier AI split the industry into three camps within days, while Deloitte finds only 21% of organizations have mature agentic AI governance.
A developer found a 313MB encrypted archive of his project's Git history queued for Alibaba's cloud storage after 564 failed upload attempts.
Four consumers filed a class action in California on 18 September, alleging the labs agreed to slow capability advancement and citing their own public statements as evidence.
Runway showed research on streaming AI video as users describe it, aiming to cut waiting time and GPU costs; no release timeline announced.
Prompt injection occurs when untrusted content supplies instructions that compete with the agent's intended task, causing the model to confuse data with authority.
Meta's personal AI agent Muse was downloaded over 900,000 times in its first week, but testing shows the app pushes users to hand over more personal data.