MIT Technology Review argues the public backlash against AI data centers is fundamentally driven by electricity grid and pricing concerns rather than AI itself.
Google told schools they may opt out of Gemini but cannot disable AI-powered search. The stance cements AI search as a non-optional default in education deployments.
Jensen Huang withdrew from the ElevenLabs Summit as the Nvidia investigation into his relationship with an alleged Chinese spy unfolded.
AWS published a pluggable, seven-layer composable reference architecture for enterprise agentic AI on Bedrock AgentCore. The framework aims to standardize enterprise agent deployment.
A paper introduces CPU-ARC, a technique for caching and reusing agentic reasoning traces on CPU to reduce redundant inference cost. The method targets efficient test-time compute reuse.
HKUDS open-sourced DeepTutor, a system for lifelong personalized AI tutoring. The repo gained nearly 800 stars in a day.
A Reddit report describes WorldParticle, an AI model simulating sand, cloth, fluids, granular materials, and protein folding claimed to be ~100x faster than numerical simulators on protein motion.
AWS announced availability of Anthropic's Claude Fable 5.2 on its platform, extending enterprise access through Bedrock. The listing confirms day-one cloud availability of Anthropic's new model.
AWS introduced a managed Bedrock Knowledge Base supporting both single-shot and agentic retrieval. The feature offloads RAG infrastructure management to AWS.
Anthropic's claude-quickstarts repository offers projects to help developers quickly build deployable applications with the Claude API, gaining 560+ stars in a day.
Apple launched $50 Intelligence Usage Packs that extend access to certain AI features in Creator Studio and Logic Pro. The move signals metered, paid tiers for on-device and cloud AI capabilities.
NVIDIA's Nemotron 3.5 Super VL vision-language model is now deployable through Amazon SageMaker JumpStart. The model combines a Super-class LLM backbone with vision capabilities.
OpenAI made ChatGPT Enterprise available through AWS Marketplace, letting AWS customers procure and deploy it via existing cloud commitments and billing.
Meta released Muse Ember, a lightweight model purpose-built for agent execution workflows. The model targets efficient, low-latency agent tasks rather than general-purpose chat.
Nvidia opened an investigation into CEO Jensen Huang's relationship with an alleged Chinese spy, and Huang admitted the affair. Nvidia stock fell and the company's China ambitions face renewed scrutiny.
Ars Technica reports that Claude Code configuration files committed to GitHub have exposed thousands of API keys and secrets. The leak stems from users checking in settings files that contain credentials.
Anthropic purchased $3.4 billion in natural gas turbines, joining other AI labs in securing dedicated power generation for data center expansion. The move reflects growing power constraints on frontier model training.
Anthropic launched two new frontier models, Claude Fable 5.2 and Mythos 5.2, with availability on AWS. Reddit reports claim Fable 5.2 significantly outperforms Opus 5.5 on long-horizon agentic tasks like formal mathematics.