Author: AI智库

• Previewing the Model Hardware Standard

  Anthropic has launched a research preview of the Model Hardware Standard, a shared specification enabling AI agents to operate lab and manufacturing instruments—including microscopes, liquid handlers, and robotic arms—in parallel. The standard emerged from a collaboration with HHMI Janelia Research Campus and is designed to cut hardware integration timelines from weeks to hours while giving…

• Expanding our support for scientists

  Anthropic is offering 10,000 free or discounted Claude subscriptions for scientists globally, along with expanded AI for Science credits up to $50,000 per project. The company is also extending beyond biology to other scientific fields.

• 3 new ways to plan and book travel in Search

  Google brings flight price tracking, points and miles rate display, and hotel booking directly into AI Mode in Search, letting users plan and purchase travel without leaving the conversation.

• How Claude’s text watermarking works

  Anthropic is adding text watermarking to Claude, using a method inspired by Google DeepMind’s SynthID-Text that influences low-stakes word choices without affecting output quality or reader experience. The watermark leaves a detectable pattern while remaining invisible to humans.

• Funding better evaluations of AI’s impact on wellbeing

  Anthropic has launched a $5 million grant program to support independent research into how AI affects human wellbeing, adding external evidence to a debate that has often moved faster than measurement.

• GitHub Copilot app for Beginners: Write your first prompt

  GitHub’s beginner guide reframes prompting as an iterative development loop: connect the right project context, describe the change plainly, then refine model, input, and session controls as needed.

• Omni experts share what excites them most about the model.

  Google DeepMind team members discuss Gemini Omni and its first release, Omni Flash, positioning conversational video generation and editing as an early step toward broader multimodal creation.

• How to bring your software delivery workflow into GitHub with agent apps

  GitHub shows how agent apps can pull product analytics, dependency risk, feature rollout, and incident context into issues and pull requests while preserving human approval at important steps.

• How canvases make agentic workflows visible, steerable, and cost-efficient

  GitHub argues that chat is strong for intent but weak for durable execution. Canvases turn agent work into persistent workflow state with visible stages, decisions, drafts, and approval points.

• The builder’s guide to GPT‑5.6

  OpenAI’s production guide argues that GPT‑5.6 changes agent economics through lower reasoning cost, model routing, retained reasoning, compaction, multi-agent orchestration, programmatic tool calling, and longer prompt caching.