Get the TERNZ AI weekly newsletter: the week's top AI travel tools and AI news, by email.

Moonshots EP #305 Reaction: Claude-Designed Scorpion Toxin Binders

In my reaction to Moonshots EP #305, the story that stopped me was a scorpion-venom breakthrough: Anthropic's Claude designed protein binders against scorpion toxins, and 17 of 211 lab-tested designs bound their target. It is early research, not yet a working antivenom, but the designs were released publicly.

TERNZ AI · My reaction
Moonshots vs. All-In: The AI Story That Stopped Me

My main takes

  • AI is now designing physical biology, not just text: Claude spent about 64 hours designing candidate proteins against five scorpion neurotoxins.
  • The result is a proof of concept. 17 of 211 lab-tested designs met the binding criteria, and the strongest bound a Mexican scorpion toxin at single-digit nanomolar strength. Neutralization has not been shown in cells or animals.
  • Releasing all 211 designs publicly means other labs can test, improve or discard them. That openness is what could turn a promising result into a real antivenom.
  • Scorpion stings are a serious danger in many parts of the world, so a faster, cheaper path to antivenom would matter far beyond the headline.

The podcast I’m reacting to

Sources for this story

All reactions →   Subscribe on YouTube →

Get the weekly TERNZ AI newsletter

One short email a week with my reaction video and the top AI stories.