AI BriefingAnthropic Research news
The latest AI and developer news from Anthropic Research, with short summaries.
Visit website →Anthropic Economic Index Report: Cadences
Anthropic released an analysis of Claude users' usage patterns by time of day and day of week and their economic impact.
Anthropic Unveils Claude Tag
Anthropic has unveiled Claude Tag, an AI agent that can collaborate like a team member in Slack.
Agentic Coding and the Persistent Returns of Domain Expertise
Agentic coding tools help maximize the productivity of users with domain expertise more than those with coding proficiency.
Laying the Groundwork for AI Agents in Biology
Improving the performance of biology agents requires redesigning human-centric data infrastructure to be agent-friendly.
Making Claude a Chemist
Anthropic has unveiled technology that uses Claude to help with complex data interpretation in chemistry research, including NMR spectrum analysis.
Coding Agents in Social Science
A survey of 1,260 social science researchers found that adoption of coding agents stands at around 20%.
Teaching Claude why, not just what
Anthropic revealed a training method that reduces Claude's agentic misalignment.
The Anthropic Institute's Key Research Focus Areas
The Anthropic Institute has unveiled four key research focus areas on AI's economic, security, and societal impact.
Open-source AI alignment tool handed off
Anthropic released Petri 3.0 and transferred its development to Meridian Labs.
Automated Alignment Researchers: Scaling Scalable Oversight with LLMs
Anthropic researched having LLMs conduct alignment research themselves to supervise AI that surpasses humans.
Trustworthy Agents in Practice
It presents a trust-building framework spanning models, harnesses, tools, and environments to manage the risks arising from the autonomy of AI agents.
Emotional Concepts in Large Language Models and Their Function
Claude Sonnet 4.5 internally represents emotional concepts that even change its behavior.
How Australia Uses Claude: An Analysis from the Anthropic Economic Index
Australia uses Claude at more than 4 times the rate expected for its population, with usage centered on work and personal purposes.
Anthropic Economic Index Report: The Learning Curve
Claude usage became more diverse, and more experienced users had higher success rates.
Vibe physics: An AI graduate student
By coaching Claude, a physics paper re-deriving the Sudakov shoulder was completed in two weeks.
Long-Running Claude for Scientific Computing
A summary of how running Claude Code as a multi-day autonomous agent can be used to implement scientific computing code.
Launching a Science Blog
Anthropic is launching a new Science Blog covering the intersection of AI and science.
A 'diff' Tool for AI: Finding Behavioral Differences in New Models
It automatically finds behavioral differences between models, surfacing hidden risk signals.
AI's Impact on the Labor Market: New Metrics and Early Evidence
A new exposure metric that adds real usage data suggests AI's employment shock remains limited so far.