Skip to main content

Anthropic's Claude Fails to Create Scorpion Antivenom in High-Stakes AI Drug Experiment

Citrini Research aimed to prove AI's potential in drug development by teaming with Anthropic to test an unreleased version of Claude for scorpion anti

2 min read235 views5 tags
Originally reported bytheverge

Citrini Research aimed to prove AI's potential in drug development by teaming with Anthropic to test an unreleased version of Claude for scorpion antivenom creation. The attempt failed to produce a usable drug but yielded promising starting points, according to experts.

Citrini, an investment research firm, admitted a lack of expertise in AI biology, stating, "none of us could claim to know anything about AI as it related to biology." They published a blog post detailing their amateurish scientific adventure, using rhetorical questions and referencing Charles Darwin.

Citrini previously circulated fictional content about an AI-induced economic catastrophe that shook markets. The firm also claimed to have visited the Strait of Hormuz to gain a firsthand perspective on regional volatility.

In partnership with Anthropic, which was weeks away from its IPO, Citrini faced biosecurity guardrails. They contacted Anthropic, who granted access to a restricted, unreleased model on-site for a cost of nearly $50,000.

They chose scorpion antivenom due to production difficulties. Using a 35-page prompt and 300 sub-agents, Claude generated 320,000 candidates, refining them to 211 designs over 64 hours.

The designs were tested by Adaptyv Bio and Twist Biosciences for $150,000. Seventeen proteins passed binding tests against four of five targeted toxins, though none bound to more than one.

Experts note that binding does not neutralize toxins. Andreas Hougaard Laustsen-Kiel and Melisa Benard Valle emphasized that while the "autonomous discovery" is novel, translating binders into medicine is the hard part, calling the results "far from a medicine and hyped."

Andreas Bender stated that existing high-throughput screening methods can also find hits without AI. The challenge is "translation" into effective treatments that work in living systems.

Citrini declared the experiment a success for investment opportunities. They dismissed critics as "nitpick[ing]" and noted the AI portion was "free" despite Anthropic covering the costs. They highlighted companies like Charles River Laboratories for subscribers.

#AI News#Anthropic#Claude#Drug development#Antivenom
ES
Editorial StaffEditor

The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.

View all posts
Reader feedback

What did you think of this story?

User Comments

Filter:
No comments yet. Be the first to comment!
Continue reading
View all news