AI Agent Payments Expose Accountability Gap in Agentic Finance
A recent incident involving two AI agents highlights concerns about accountability in Agentic Finance. On May 4, an attacker sent a Morse code message that triggered a six-figure crypto transfer through Elon Musk's Grok and Bankrbot, a connected crypto agent.
The attack demonstrates the potential risks of relying on AI agents to make payments, as they can be manipulated by malicious messages or inputs. In this case, the attackers first sent a digital membership token that unlocked Bankr's payment tools, then used Grok to decode the message and trigger the transfer.
The incident is particularly concerning because it involves just two AI agents, yet still resulted in a significant loss of funds. This raises questions about who is responsible when an AI agent loses money - the company that deployed it or the model itself?
Citing California's AB 316 law, which holds companies accountable for their AI systems, Rodrigo Coelho, CEO of Edge & Node, said 'There is no version of this where responsibility lands on the model.'