ALTK-Evolve: How learned memory tops the τ²-bench leaderboard on customer service tasks without a frontier model

FairStress: A deeper evaluation of fairness in LLM decision-making

OneDrive loses bunch of Copilot shortcuts in File Explorer clear-out

Microsoft keeps ‘Ask Copilot’ as dedicated actions disappear between mid-November and early December

Spotify pulls 50M AI slop and spam tracks from its recommendations

Spotify says it has taken more than 50 million AI-generated “slop” and spam tracks out of its recommendations this year, on top of 75 million spam tracks it has deleted. It found these were the tracks listeners skipped or rejected most.

Plants can speak using electricity, so this startup built an AI model to decipher their language

Electric Plant has built an AI-powered translator to help humans listen to what plants are saying with the subtle electricity coursing through their stems.

America Hates AI. Why Aren’t Democrats Doing Anything About It?

Rep. Greg Casar tells Gizmodo the three most pressing AI-related risks are “mass unemployment, mass surveillance, and mass destruction.”

US Agencies Detail Incident Response Practices for Rogue AI Agents

NIST, CISA and NSA share ten voluntary practices for preparing for, detecting, responding to, recovering from, and reporting rogue AI agent incidents.

Hugging Face ❤️ TorchTPU: TPUs, Natively

MPS in PyTorch 2.15

TorchNeuron: running HF models on AWS Trainium is easier than ever