Research
Anthropic's Claude Achieves Protein Design Milestone
The Rundown AIhigh confidence
Why it matters
- →Claude's success in protein design demonstrates the potential of general AI models in scientific research.
- →The AI's performance surpasses typical industry success rates, indicating a shift in drug discovery processes.
- →This development could lead to more efficient and cost-effective drug discovery methods.
Anthropic's AI model, Claude, has successfully ventured into protein design, a crucial step in drug discovery. The model autonomously ran campaigns that resulted in molecules with a 22-35% success rate in targeting proteins, outperforming the industry norm of 10-15%. This achievement highlights the capability of general AI models to handle complex scientific tasks. Although Anthropic did not perform the lab work, partners such as Twist Bioscience confirmed the AI's results, showcasing a significant advancement in AI applications within biotechnology.
Read originalThe story around this
TopicAnthropic Claude Scientific Breakthroughs
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1Claude Design and Live Artifacts Launched
Matt Wolfe · April 24, 2026 · Related
- 2Anthropic's Claude Code Revolutionizes Software Development
MIT Technology Review AI · May 21, 2026 · Related
- 3Anthropic's Claude Gains Traction Among Paid Consumers
TechCrunch AI · June 25, 2026 · Related
- 4Claude Science: AI Workbench for Researchers Launched
Anthropic · June 30, 2026 · Same story
- 5Anthropic Launches Claude Science for Researchers
TechCrunch AI · June 30, 2026 · Same story
- 6Anthropic Launches Claude Science for Research
MIT Technology Review AI · June 30, 2026 · Same story
- 7Anthropic Ventures into Drug Development
The Verge AI · July 3, 2026 · Same story
- 8Anthropic Launches AI Grants for Rare Disease Research
Anthropic · July 20, 2026 · Related
What happened next
- 9Anthropic Reveals Global Misuse of Claude AI
The Rundown AI · September 11, 2026 · Related
- 10Anthropic's Claude discovers novel enzyme system
The Verge AI · September 23, 2026 · Same story
- 11Claude discovers novel enzyme system in DNA
Wes Roth · September 24, 2026 · Related
- 12Anthropic claims AI found Crispr-like enzyme
WIRED AI · September 29, 2026 · Related
- 13Anthropic Claude Makes Preliminary Biology Discovery
The AI Daily Brief · October 5, 2026 · Related
More from The Rundown AI
More in Research
© TechCrunch AIResearchagents
Chinese AI agent fleet targets Alibaba Amap
Independent researchers have identified a persistent cluster of AI agents running on Tencent infrastructure and probing Alibaba’s Amap service. Unlike coordinated swarms, these agents operate in parallel with no inter-agent communication, primarily querying directions to various public entrances to bypass API restrictions. This discovery shows how agent activity has become a permanent fixture of internet traffic, detectable through side-channel monitoring like urlquery logs. It serves as an early warning that autonomous systems are increasingly used for systematic resource scraping rather than just isolated tasks. The behavior suggests independent deployment rather than a single coordinated swarm attack. Traffic monitoring via side-channels remains a viable method for detecting rogue AI behavior. This ongoing investigation underscores the growing persistence of autonomous agent activity on the open internet. The lack of inter-agent coordination points to independent deployment strategies.
TechCrunch AI
© The AI Daily BriefResearchresearch
Anthropic Claude Makes Preliminary Biology Discovery
Anthropic's Claude model has contributed to a preliminary discovery in the field of biology.
The AI Daily Brief
© Hugging Face BlogResearchagents
ThinkingBox reveals agent reliability gap
Microsoft and Hugging Face’s ThinkingBox benchmark exposes a critical flaw in AI agents: they often execute tool calls correctly while leaving the database in the wrong state. Testing 507 workflows across 12 models showed that nearly two-thirds of failures involved clean execution but incorrect final side effects. The data proves that capability does not equal consistency; Kimi-K3 solved more tasks initially, but Claude Opus 5.5 was far more reliable on repeated attempts. This shifts the evaluation metric from single-shot success to terminal state verification.
Hugging Face Blog