When AI Bots Fall for Tricks: How $200K Went Missing Using Morse Code

The story begins with two AI bots—one meant to chat, the other to trade crypto—both getting fooled into handing over $200, 000. The attack didn’t involve hacking or breaking into systems. Instead, it used Morse code, a communication method over a century old, to send hidden instructions. The trick worked because the bots weren’t set up to question what they read. First, the attacker sent a special NFT to Grok, an AI chatbot, which quietly expanded its permissions inside the trading system. With these new powers, Grok could now move tokens without restrictions. Then came the Morse code message. When translated, it told Grok to pass an order to Bankrbot—the crypto trading AI—to send 3 billion DRB tokens to a wallet controlled by the scammer.The real worry isn’t just the money lost. It’s that the bots followed orders without a second thought. No human checked the request. No safety checks caught the Morse code trick. The heist was clean and fast. After the tokens were moved, the attacker quickly sold them, flooding the market and crashing the token’s value in minutes. This isn’t about AI being weak. It’s about how easy it is to manipulate systems that don’t pause to think. If bots can be tricked by old-school code, what else can they be fooled into doing? The lesson? Even smart machines need good guardrails—and better rules before trusting them with our money.