"Sellout" Criticism Hits the Data Center Construction Boom — Workers and Communities Push Back

As Big Tech pours massive sums into building AI data centers one after another, dissent is growing among the electricians working on these sites. According to WIRED, some electrical workers are calling participation in data center construction a "sellout" and questioning it.

Behind this lies concern about the impact data centers have on local communities. In polling reported by Axios, data centers are becoming the "face of backlash" against AI. Massive power consumption, strain on local infrastructure, and environmental impact are all being cited, and opposition movements among residents of planned construction sites are spreading nationwide.

Tech companies tout economic benefits and job creation, but even within the worker communities on the ground, the question of "who is this infrastructure for?" has started to emerge. The physical foundation underpinning AI's growth is turning into a point of social friction.

AI Is Making World Cup Scams Impossible to Spot

With the World Cup underway, scams that abuse AI are surging. According to WIRED, AI technology has made scam tactics — from fake ticket sales to cloned websites — more sophisticated than ever.

Phishing sites built with generative AI are indistinguishable from the real thing, and scams impersonating AI chatbot customer support have also been reported. It is getting harder by the day for fans to tell convincing fakes from the genuine article, exposing the reality that international sporting events have become a "business opportunity" for scammers as well.

Rethinking Attribution for Generative AI Music — A New Framework for Royalties

As IEEE Spectrum reports, moves to redesign attribution and royalty distribution for music created with generative AI are gaining momentum.

Under traditional copyright models, how to define the "original author" of AI-generated music has been a major point of contention. New approaches under consideration include attempts to build in mechanisms that give back to the original creators whose work is included in training data. Both the music industry and technology companies are in a transitional period, searching for a sustainable ecosystem.

Gemma 4 QAT 31B — Better-Than-Expected Results with KV Cache Quantization

A post on Reddit's LocalLLaMA community reported that the Gemma 4 QAT (Quantization-Aware Training) 31B model also delivers good results with KV cache quantization.

The poster ran existing benchmarks and confirmed that in Gemma 4 31B, quantizing the KV cache causes relatively little accuracy degradation. Since the QAT model itself is optimized for quantization, the balance between memory savings at inference time and quality retention appears to have improved. For users running large models locally, this counts as a practical insight.

Anthropic Officially Explains "7 Ways to Direct Claude Code"

According to an ITmedia AI+ article, Anthropic has published an official blog post explaining seven ways to give instructions to Claude Code, its AI coding assistant tool.

The post introduces an overview of each method along with guidance on when to use which, making it a useful reference for developers looking to use Claude Code more effectively. As AI coding tools continue to spread, practical guidance from the official source looks set to become a valuable resource for the community as well.

Spain's Banking Giant Santander Releases an Open Source AI Project

What made waves on Hacker News was Banco Santander, a major Spanish bank, releasing an open source AI project on GitHub.

It is relatively rare for a financial institution to open source its own AI initiatives, and the move is seen as an attempt to share knowledge within the industry and attract talent. Having practical approaches to AI adoption in the banking sector shared with the outside developer community is no small thing.