Commonwealth Bank of Australia (CommBank) has started testing CommBank Companion, an AI-powered conversational banking experience embedded inside its mobile app, as the financial institution expands its push into GenAI while emphasizing governance and customer safeguards.
CommBank is testing the new capability, which is designed to help retail and business users manage their finances, with employees and selected business banking customers.
Given growing industry and regulatory concerns around AI security in financial services, including fraud, data privacy, hallucinations, identity risks and the potential misuse of GenAI in customer interactions, CommBank is positioning security investment as a core component of its AI rollout strategy.
The bank said CommBank Companion operates within its existing secure banking environment and is supported by around $1BN in security investment, “building on CommBank’s ongoing focus on protecting customers from scams and fraud, including real-time safeguards embedded directly into the app.”
That emphasis reflects increasing pressure on banks to demonstrate that AI-enabled customer experiences can meet the same trust and compliance standards expected of traditional banking systems.
Angus Sullivan, Group Executive Retail Banking Services at CommBank, said:
“Trust is critical in banking. That’s why we’re taking a deliberate and staged approach to CommBank Companion, with strong governance, testing and customer safeguards embedded from day one.”
The bank’s engineering and data science teams designed and built CommBank Companion in-house. Contextualized in customer data for transaction-related questions, it understands customer spending and savings, using AI techniques to deliver fast, relevant responses. Commbank emphasized that there are “clear guardrails and controls” in place informing the app’s behavior.
Commbank Positions AI as a Decision-Support Tool, Not a Decision-Maker
The conversational interface in the app brings together live spending and saving data in one place to deliver relevant information, helping the customer make better-informed financial decisions.
During testing, CommBank Companion can respond to customer questions in everyday language using transaction and savings data to provide personalized insights on budgeting, subscriptions, savings goals, borrowing power, home valuations, loan comparisons, deposits and repayment schedules directly within the app.
The AI assistant offers business customers real-time visibility into cash flow, including historical trends, projections and suggested next steps when opportunities or financial pressures emerge. CommBank said it will develop the Companion over time to help businesses model hiring or expansion scenarios, identify emerging cashflow pressures such as supplier or energy increases, benchmark performance against similar businesses and manage servicing tasks including statement requests and merchant terminal support, with escalation to human customer support where required.
Mike Vacy‑Lyle, Group Executive, Business Banking at CommBank, said that business owners often have access to data, but struggle to use that intelligence to support decision‑making.
“CommBank Companion is designed to bring clarity to some of that complexity—summarizing what’s happening in the finance of the business in plain language. As the experience is refined, it will highlight when attention may be needed to help business owners sense‑check financial decisions before they act.”
The bank stressed the platform does not provide regulated financial advice or credit assistance on regulated products, nor does it make decisions on behalf of customers. Instead, the assistant is designed to support decision-making while maintaining human oversight and customer control.
That positioning reflects growing concern around hallucinations, inaccurate recommendations, prompt injection attacks and privacy exposure in AI-driven financial interactions. Financial institutions around the world are coming under increasing scrutiny around how they deploy AI in customer-facing environments, particularly in areas involving sensitive financial data, fraud prevention, explainability and regulated financial guidance.
