News & analysis, rated.
Breaking AI developments, in-depth guides, real-world case studies, and analysis — each one rated so you know what matters.
Gen Z workers over-relying on AI at work—here's the career cost
Fortune reports that younger workers are depending too heavily on AI tools without building core skills. The trend could undermine long-term career prospects and employability.
Gen Z hiring problem isn't new, WeWork and Upwork CEOs say
WeWork and Upwork leaders confirm Gen Z workforce challenges, but the real story is how long these gaps have persisted in corporate recruiting.
Cornerstone bets $1B learning platform on AI agents and skills graphs
The largest corporate learning vendor is pivoting from SCORM-based course authoring to AI-native content generation and personalized employee development agents. Early customers report skills inference and retention gains.
Google's Manyika: Job losses won't hit as fast as AI labs predict
James Manyika, Google's AI strategy lead, pushes back on Microsoft and Anthropic predictions of mass white-collar job losses within 18 months. He cites a decade of labor research showing tasks automate faster than whole jobs do.
Stop Grading Agents on Model Scores—Measure Real Task Success
NVIDIA outlines five practical methods to evaluate AI agents on actual workflows, not benchmark points. Learn why trajectory logging and tool call precision matter more than MMLU scores.
NVIDIA Verified Agent Skills: Scan, Sign, Deploy Agents Safely
NVIDIA is publishing a catalog of cryptographically signed agent skills with scanned vulnerabilities, skill cards, and provenance tracking. Here's what developers need to verify before deployment.
Six rerankers beat their size class on MTEB, all under Apache 2.0
Hugging Face released six cross-encoder rerankers (17M to 1B params) trained via distillation on ModernBERT encoders. Full training recipe and data included; benchmarked on MTEB retrieval tasks.
OlmoEarth v1.1 cuts satellite model costs 3x with shorter tokens
Allen Institute released OlmoEarth v1.1, a satellite imagery model family that slashes compute costs by up to 3x while matching v1 performance. The trick: merging multi-resolution tokens without sacrificing accuracy.
DeepMind's Co-Scientist finds genetic factors that reverse aging in human cells
DeepMind's AI tool scanned tens of thousands of aging research papers and proposed 20+ novel genetic factors to test. Lab validation confirmed several candidates successfully rejuvenate cells.
KPMG deploys Claude across 276,000 staff in enterprise alliance
KPMG is integrating Anthropic's Claude into core business operations for its entire global workforce. The deal marks a major enterprise adoption, but specifics on deployment scope and use cases remain undisclosed.
Anthropic expands frontier AI debate beyond safety
Anthropic publishes new framework for discussing frontier AI risks and governance. The company aims to broaden industry conversation beyond technical safety to policy and societal impact.
OpenAI Adds Content Credentials and SynthID to Flag AI-Generated Media
OpenAI is integrating Content Credentials and expanding SynthID watermarking to help users identify AI-generated images and text. The company also released a verification tool for detecting synthetic media.