• •ZDNet reports that 'agentic payments' are enabling AI agents to initiate and execute transactions autonomously on behalf of users.
• •The process currently leverages existing APIs and OAuth integrations, allowing agents to act within defined financial sandboxes or specific banking interfaces.
• •A significant unresolved question remains regarding liability: there is currently no industry-standard framework for who is responsible when an AI agent makes an unauthorized or erroneous transaction.

AI agents are increasingly being equipped to handle financial transactions, shifting the role of software from passive information gathering to active fund management. Unlike standard automation scripts, these agentic systems utilize large language models to interpret intent and execute payments in real-time across various banking platforms. However, the lack of robust security protocols and legal frameworks surrounding autonomous financial decisions creates substantial operational risk. Widespread adoption will likely depend on whether financial institutions can implement 'human-in-the-loop' verification systems that scale without negating the speed these agents promise.