OpenAI says two API settings tripled GPT-5.6 scores on ARC-AGI-3, underscoring how inference configuration can reshape model performance.

A reported Hugging Face breach tied to OpenAI testing has renewed debate over AI agent sandboxing, responsibility, and enterprise risk controls.

Coverage from ZDNET and The Tech Buzz points to a sharper split between open-weight and closed AI models, with major stakes for cost, control, and competition.

Block’s Goose is gaining attention as a free, local alternative to Claude Code, highlighting cost, privacy and control trade-offs in AI coding tools.

Encore AI has raised a $30 million Series A to build voice and text agents trained on customer calls, a bet on workflow-specific enterprise AI.

SK Telecom introduced a second A.X K2 Korean-language model, signaling a deeper push into domestic enterprise AI with stronger reasoning claims.

Anthropic says Claude Mythos found flaws in HAWK and reduced-round AES, signaling AI may accelerate cryptanalysis without threatening current systems.

Coverage in The Guardian and Cybersecurity Insiders highlights a new push to measure AI agent behavior before deployment as enterprises weigh safety risks.

Microsoft says a new AI security model outperformed Mythos on a benchmark, signaling a push into specialized cyber AI as buyers demand proof over hype.

A Communications of the ACM article proposes a framework for evaluating openness in foundation models, a key issue for enterprise AI buyers and builders.

NVIDIA has open-sourced a GPU-accelerated medical physics simulation framework for healthcare robotics, aiming to speed robot training and testing.