What happened after 2,000 people tried to hack my AI assistant
Fernando Irarrázaval ran a challenge on hackmyclaw.com where 2,000 people attempted to hack his OpenClaw AI assistant via email. After 6,000 attempts, $500 in token spend, and a Google account suspension, nobody leaked the secret. Simon Willison notes that frontier models are becoming harder to inject, but warns against deploying production systems where injection could cause irreversible damage.
Even 6,000 failed prompt injection attempts don't guarantee security; production systems need robust defenses.


