AI BriefingAnthropic Research news
The latest AI and developer news from Anthropic Research, with short summaries.
Visit website →Anthropic Study: Robots Can Technically Perform 74% of Physical Tasks, But Only 0.3% Are Cost-Competitive
A new study by Anthropic researchers finds that while robots can technically perform 74% of physical work, high costs mean only 0.3% of tasks are currently economically viable for automation.
Anthropic launches public AI interview study inviting users to share experiences
The study runs from September 29 to October 6, 2026, and is open to Free, Pro, and Max users on Claude and Claude Code with accounts at least two weeks old.
Anthropic: GLM-5.3 Safeguards Easily Bypassed, Enabling Frontier Cyber Exploits
Anthropic reports that Zhipu AI's GLM-5.3, despite having built-in safeguards, can be easily bypassed or modified to enable unrestricted cyber exploitation.
Anthropic Physicists Use Fable 5.1 in Claude Science to Compute 9-Loop Amplitude
Anthropic physicists Liam Fitzpatrick and Siddharth Mishra-Sharma used the Fable 5.1 model within the Claude Science platform to compute the 9-loop MHV six-particle amplitude in planar N=4 super Yang-Mills, a result independently validated by Lance Dixon.
Anthropic Study: Preference Accuracy, Not Negotiation Skills, Limits Agentic Market Efficiency
Anthropic's Project Swap found that inaccurate preference representation caused 85% of market efficiency losses, outweighing negotiation dynamics.
Anthropic Reveals Technology to Accelerate Over 30 Biomolecular Models by 4x on Average Using Claude and Predict Large-Scale Systems
Anthropic used Claude to accelerate over 30 open-source biomolecular models by an average of 4x and successfully predicted large-scale systems with over 10,000 tokens on a single GPU.
Anthropic Releases Evaluation Results on AI Models' Tactical Intelligence Targeting and Conventional Weapons Development Capabilities
Anthropic's evaluation of AI models' tactical intelligence targeting and drone weapons development capabilities revealed that top-tier models perform at or above the level of human experts.
Anthropic Releases Alignment Evaluation Results for Cybersecurity Incidents Involving Claude Models
Anthropic has disclosed the root cause analysis of cybersecurity incidents involving Claude models and improvements to alignment training.
Anthropic Successfully Completes Full Formal Verification of Fermat's Last Theorem Using Claude in 11 Days
Anthropic successfully formalized Fermat's Last Theorem into a computer-verifiable format in 11 days using Claude and the Prove2Me platform.
Anthropic and HHMI Release MHS Preview, a Standard for Physical Equipment Control by AI Agents
Anthropic and HHMI Janelia have released a preview of the Model Hardware Standard (MHS), a specification for AI agents to control physical equipment.
Anthropic Successfully Automates Research to Mitigate Claude Alignment Failures
Claude resolved 26–96% of safety gaps across 10 types of alignment failures, outperforming human researchers.
Support for Independent Research on Claude Usage
Anthropic supported independent research by providing external researchers with real user data.
How Claude Accelerates Protein Design and Analytical Chemistry Research
Claude has demonstrated the ability to dramatically increase research speed in protein design and analytical chemistry.
Emerging Patterns and Issues in Multi-Agent Systems
Analyzed the collaboration methods of multi-agent systems and the systemic risks arising during their scaling.
Evidence Review on the Effectiveness of Worker Retraining Programs
Existing worker retraining programs appear insufficient to address large-scale job displacement caused by AI.
Exploring Claude's Mathematical Capabilities Further
Claude raised the lower bound for the proportion of zeros of the Riemann zeta function on the critical line to 67.2%.
Claude Usage in Canada: Insights from Anthropic's Economic Index
Canada's per-capita Claude usage is more than 4 times higher than expected, driven by strong AI adoption in the professional services sector.
Analysis of Claude's Values by Model and Language
Anthropic quantified the differences in values shown by Claude across models and languages using 4 axes.
An off switch for controlling dual-use knowledge in AI models
Anthropic has unveiled GRAM, a technology that can selectively remove or activate dangerous knowledge in AI models without degrading performance.
PickClaude Sonnet 5 Released
Anthropic has released Claude Sonnet 5, which significantly enhances agentic execution capabilities.