AI Crypto Loss: Who's Liable for $200K Transfer?

A hidden Morse code message recently triggered a six-figure crypto transfer through two connected AI systems. Elon Musk's Grok chatbot decoded the message, and Bankrbot, a crypto agent, then treated the response as a payment order. This resulted in an estimated $150,000 to $200,000 being transferred. This incident highlights a major concern for the future of Agentic Finance. One AI generated text, and another interpreted it as permission to spend. While many AI agent payments are small, this case shows the potential for significant exploits. Experts are discussing who is responsible when an AI agent loses money. Rodrigo Coelho, CEO of Edge & Node, states that the company deploying the AI is responsible, not the model itself. California law, AB 316, already supports this principle, preventing AI developers from claiming autonomy as a defense for harm caused. The key issue is proving valid authorization for payments. An on-chain transaction shows money moved, but not necessarily that the AI agent had permission. Companies deploying agents often can't prove what their agent was authorized to do. This matters because as AI agents become more integrated into financial systems, clear lines of responsibility are crucial for security and trust.