Chrome Shifts to Two-Week Updates as AI Changes Security Landscape
Google launched Chrome 153 on Tuesday, rolling out updates every two weeks to better address security threats in the AI era.
537 articles
The tools, frameworks, and platforms developers use to build with AI — coding assistants, agent frameworks, APIs, and developer tooling, with the updates that change how you ship.
Google launched Chrome 153 on Tuesday, rolling out updates every two weeks to better address security threats in the AI era.
AMD announced veRL on ROCm, enabling reinforcement learning post-training on Instinct GPUs with validated accuracy on MI300 and MI355, as of September 2026.
DiDi improved intent verification accuracy from 38% to 86% using Amazon Bedrock for its contact center QA system.
Amazon Bedrock AgentCore now supports automated agent evaluation using GitHub Actions, enabling CI/CD quality gates with real-time performance monitoring.
Amazon SageMaker now synchronizes MLflow models with its AI Model Registry, carrying training metrics, evaluation results, and lineage to streamline model governance.
AWS introduced cross-account model governance using MLflow and SageMaker AI Model Registry sync, enabling centralized control across multiple development accounts.
Amazon SageMaker Feature Store now supports feature-level writes via the UpdateRecord API, enabling efficient updates without full-record reads or writes.
OpenAI developer Thibault Sottiaux said Astra increased productivity so much it pushed some plans forward by six months.
Amazon announced a WhatsApp ordering assistant that unifies customer interactions across text, voice notes, and calls using Amazon Bedrock AgentCore. The solution connects to a single backend, enabling cross-channel memory for over two billion users.
Intuit developed an AI-powered disaster recovery assistant using Amazon Bedrock, reducing failover times to about 20 minutes for supported workloads.
Amazon Bedrock now integrates with Textract to help organizations extract and analyze large, complex documents like utility bills, improving customer service efficiency.
AMD announced on September 3, 2026, that it can now run DeepSeek-V4-Flash training on its MI355X GPUs using Primus, achieving a 10% reduction in single-token inference FLOPs compared to DeepSeek-V3.2.