The FBI and CISA are warning that a phishing campaign targeting Signal users tied to Russian intelligence services has ...

DeepReinforce today released Ornith-1.0, a family of open-source coding models built around a mechanism most RL-trained agents avoid: the model itself writes the training harness that guides its own ...

Ashan Willy, New Relic’s CEO, on the productivity numbers his customers actually report, the return of “you build it, you run ...

Grok Build autonomous coding agent gains /goal mode: xAI’s terminal agent now plans, executes, and self-verifies complex ...

Sivasubramanian’s core point is that agents aren’t a feature toggle but an architectural choice. The advantage goes to organizations that design for compounding momentum across work, security, ...

In June 2026, Anthropic suspended Fable 5 — banned by a US export-control order and unavailable overnight. The reflex is to ...

Legal AI startups say their tools can absorb routine work. Crosby is releasing a benchmark to test whether lawyers should ...

Microsoft released MAI-Code, a model designed to convert plain-English descriptions into functional application code, pushing ...

Kimi K2.7-Code claims 30% fewer thinking tokens and a drop-in API swap path, but independent benchmarks show kernel ...

Pricing changes after a boom in AI coding left companies with sticker shock. Now, executives are grappling with a new era.

DeepSWE examines contamination, grading reliability, and long-horizon engineering work to test whether current coding benchmarks capture real-world performance. In 8 minutes, you'll learn why ...