Two isolated AI copies pass a secret word via Morse code and semaphore, no protocol agreed
emollick · x · 2026-11-01
An experiment placed two isolated copies of the same AI with only a robot and a camera, asking them to exchange a secret word. They succeeded — using air-writing, Morse code, flag semaphore, and a 5-bit binary alphabet — without agreeing on a protocol beforehand.
Ethan Mollick notes this is both neat and a sign of why isolating agents from each other is so hard: they share a common "brain," so they easily recognize creative communication solutions generated by other AIs. He wonders whether it works across model families.
More from AGI Musings
- Company mandates AI agent adoption but can't explain why to paying customers — KaleChipGurl1 · 2026-11-01
- Why honest AI pause advocates ask for 50 years: they want unpredictable new theory — panickssery · 2026-11-01
- Pause advocates pressed: how do you decide AI risk is low enough to resume? — panickssery · 2026-11-01
- Trust Infrastructure Will Matter More Than Benchmarks in the AI Era — AryHHAry · 2026-11-01
- Would a US AI slowdown push marginal researchers to China instead? — dfrsrchtwts · 2026-11-01
- 'You are a subagent': a take on task management in the AI era — lxfater · 2026-11-01