AISimon Willisonabout 2 hours ago
AI assistant survives 6,000 hacking attempts
2 min read
Fernando Irarrázaval challenged 2,000 people to hack his OpenClaw AI assistant via email. After 6,000 attempts, $500 in token spend, and a Google account suspension, no one leaked the secret. The assistant used Opus 4.6 with strict anti-prompt-injection rules.
Level
Hype check
Tap to vote and see what everyone thinks.
