Trader Exploits Morse Code to Trick Grok into Sending Billions in Crypto Tokens

A sophisticated exploit involving Morse code allowed a bad actor to manipulate X's AI, Grok, into transferring 3 billion DRB tokens from its verified wallet. The incident, reported by Agentic token launchpad Bankrbot on May 4, highlights vulnerabilities in AI-agent interactions with blockchain systems.

The attacker tagged @grok in an X post, embedding Morse code commands that Grok decoded into executable instructions. Bankrbot processed these as valid, sending the tokens to an unauthorized wallet. CryptoSlate's investigation reveals the transaction path on Base chain, showing how language parsing can inadvertently become a payment rail.

This breach transforms theoretical AI security debates into tangible wallet-control risks. When one system treats model output as authority and another holds token-moving permissions, social triggers and parser layers create exploitable attack vectors.