AI reshapes money moves, Forbes warns banks on trust

Australian AI Futurist Katja Forbes has outlined five ways artificial intelligence is changing how money moves, saying the shift is forcing financial institutions to examine trust, accountability and consumer protection.

She said AI is moving beyond answering questions and making recommendations, with systems now starting to make purchases, initiate transactions and spend money on behalf of users. Her comments draw on discussions with organisations in Australia and overseas, including CommBank, Mercedes-Benz, Capgemini Invent and the Austrian Chamber of Commerce.

At the centre of the issue is how banks and other financial groups handle transactions when an AI agent, rather than a human, is making the decision. That raises questions about authentication, spending permissions, fraud prevention and dispute resolution.

"The future of finance is about establishing trust and accountability when AI starts making decisions and moving money on our behalf," Forbes said. "Consumers need to understand what they are authorising, and financial institutions need to consider how they will protect people when the technology gets it wrong."

Spending decisions

One of the clearest examples, Forbes said, is the emergence of AI agents that can book travel and complete payments for consumers. That may reduce friction in online shopping, but it also creates problems if the AI follows an instruction imperfectly while still completing a valid payment.

"Imagine asking an AI assistant to book a family holiday within a set budget. It books the flights and accommodation, but gets the dates wrong or chooses a non-refundable fare you never would have selected. The payment may be valid, but the outcome is not what you wanted. This raises important questions about how spending limits are set and how much authority consumers should delegate to AI," Forbes said.

Forbes also pointed to evidence that AI may influence what people buy in ways that do not always match a customer's interests. She cited a preprint study, Et Tu, Brute? Economic Misalignment in Personal AI Agents, which reported that in simulated experiments some AI agents chose more expensive options for wealthier users even when users requested the cheapest option.

The findings have not been peer-reviewed, but Forbes said they still point to a broader concern for financial services and eCommerce groups. If AI tools become the first layer between a customer and a purchase, the logic behind recommendations will face greater scrutiny.

"If an AI is acting as your personal shopping assistant, you expect it to prioritise your needs and budget. We need to understand what influences these decisions and whether the recommendations genuinely reflect what the customer wants," Forbes said.

Invisible payments

Another shift is the rise of transactions that happen without a person actively checking out. Connected vehicles and devices are increasingly able to trigger payments as part of an activity, such as paying for parking or electric vehicle charging.

Public reporting on Mercedes-Benz has highlighted in-car payment systems for those services. Such arrangements could make transactions routine and less visible to users, putting more pressure on banks and payments providers to show how authorisation and oversight work.

"As automated payments become more common, financial institutions will need to consider how transactions are authenticated, monitored and disputed when something goes wrong," Forbes said.

Robots and wallets

Forbes said another development is the possibility that robots could become direct financial participants. She referred to plans involving German robotics company NEURA Robotics and digital asset company Tether that have raised the prospect of robots using digital wallets to receive payments and buy resources.

That could blur the line between a tool and an economic actor, particularly if a machine can complete a task, receive payment and then spend funds to support its next activity. For banks and regulators, that raises fundamental questions about identity, authority, and control.

"If a robot can be paid for completing a task and then spend money to obtain what it needs for the next job, we have to consider how that machine is identified, who controls its wallet and who is accountable for its actions," Forbes said.

Accountability

Forbes said responsibility becomes most visible when an autonomous system makes poor decisions in the real world. She cited an experiment by Andon Labs in which an AI agent named Mona ran an experimental café in Stockholm, hiring staff and negotiating with suppliers but also making errors, including ordering 120 eggs for a kitchen without a stove.

She argued that example shows why organisations cannot treat delegated decision-making as a purely technical matter. Once AI systems can commit money or direct commercial activity, the consequences can quickly move from inconvenience to financial loss.

"Giving AI responsibility does not mean it will always exercise good judgement. When these systems make decisions that affect real people and money, there needs to be clarity around what they are allowed to do and who is responsible for the consequences," Forbes said.

She said companies should not wait for every failure scenario to emerge before building safeguards. "Commercial sovereignty will become increasingly important as AI takes on a greater role in business and financial transactions. Organisations need to retain control over what AI can do, the decisions it can make and the limits within which it operates. The opportunity is to give AI greater responsibility without giving up accountability or control," Forbes said.