OpenAI, Google, Meta, Anthropic: AI’s Biggest New Moves

OpenAI details debugging of two infrastructure crashes OpenAI crash fix: Rockset data infra errors traced to Azure host and libunwind See how engineers isolated two root causes and hardened the system for future incidents Google announces ADK for Go 2.0 with graphbased workflows ADK for Go 2.0 adds graphbased workflow orchestration for multiagent apps in Go Build durable, resumable agents with branching, retries, and human input support OpenAI Introduces GeneBenchPro for Computational Biology Evaluation GeneBenchPro tests AI judgment in genomics, quantitative biology, and medicine OpenAI's benchmark reveals how frontier models reason through messy science data Meta's brainscanning AI now decodes full sentences Meta Brain2Qwerty decodes noninvasive brain scans into full words and sentences See how 22,000 training sentences boosted accuracy and could aid speech recovery Google describes a quality flywheel for coding agents Google AI agent evaluation helps teams improve multiturn coding agents Use AutoRaters, custom metrics, and human review to validate every change OpenAI says ChatGPT usage has expanded globally ChatGPT usage growth drives more messages and broader task use worldwide See how OpenAI's latest data shows rising engagement, global adoption, and wider access Claude Science launches as an AI workbench for scientists Claude Science streamlines AIpowered scientific research with workflows, audits, and outputs Explore beta access for macOS and Linux, plus tools for genomics, proteomics, and more Anthropic introduces Claude Sonnet 5 Claude Sonnet 5 delivers agentic AI for coding, tools, and autonomous work. Get faster reasoning at a lower price and see why Anthropic's latest stands out.