This Week in AI

A self-spreading npm malware campaign highlights how quickly modern supply chain attacks can scale once a package lands in the wrong dependency graph.

Google AI’s STATIC framework puts sparse matrix techniques to work for constrained decoding in LLM-based generative retrieval, with the headline claim of 948x faster decoding.

A former Olympic skier and CU Buffs standout is using AI in sports to reduce missed calls, a case study in how real-time decision support is becoming a mainstream edge pattern.

AI slop in football captures a broader security problem: synthetic content can degrade trust faster than any single breach.

A proposal for AI regulations underscores the direction of travel, with governance expectations hardening, and “we are experimenting” will no longer be a sufficient posture.

AI in skincare is a window into the next enterprise playbook for personalization, combining data capture, recommendations, and continuous optimization across products and services.