Commonwealth Bank is handing ChatGPT Enterprise to nearly 50,000 employees, one of the largest workforce-wide artificial intelligence deployments attempted by any Australian organisation and among the biggest in global financial services.
The bank announced the rollout in December, framing generative AI as a core capability across the business rather than a contained pilot. In a note published on OpenAI’s site, the company said CBA set out to make AI “a core activity” for the whole workforce, pairing broad access to ChatGPT with training, connectors to internal systems and leadership role-modelling.
The scale is the headline. Australia’s largest bank employs roughly 48,000 to 50,000 people and serves about 17 million customers. Putting a generative AI assistant in front of nearly all of them, at once, is a markedly different bet from the department-by-department trials that have defined most corporate AI adoption to date.
From everyday tasks to AI agents
CBA’s stated plan runs in two phases. The first embeds AI in day-to-day work: summarising conversations, drafting compliant replies, translating, adjusting tone and pulling actions out of meetings. The second moves toward what OpenAI calls “agent-powered” use cases in higher-stakes moments, chiefly customer service and fraud and scam response.
The bank is trying to build fluency deliberately rather than hoping usage spreads on its own. According to OpenAI’s write-up, that has meant internal forums, daily hands-on tasks and structured experiments, with executives modelling the behaviour they want to see. Analysis of the rollout by consultancy Gend describes a governance layer around it, including role-based access, data-loss-prevention policies, audit logging and human-in-the-loop approval before AI-drafted responses reach customers.
The fraud angle is where the partnership goes deepest. Engineers from CBA and OpenAI are collaborating on generative AI solutions to strengthen scam and fraud detection and deliver more personalised services, a joint-engineering arrangement that goes well beyond buying software seats. It lands at a pointed moment: scam losses remain a running sore for Australian banks and their customers, and AI-generated scams, including fake news sites impersonating the ABC and cloning executives, have grown more convincing.
A national AI push, and a workforce question
The deployment sits inside OpenAI’s wider Australian expansion. Through its OpenAI for Australia programme, the company has opened a Sydney office and struck partnerships spanning Atlassian, Canva, UNSW and Coles, alongside a nationwide OpenAI Academy skills push and sovereign-infrastructure plans with data-centre operator NEXTDC. CBA is the flagship of that effort, the largest single institution to commit its whole workforce.
Chief executive Matt Comyn has cast the bank’s AI strategy in national terms. Writing in the bank’s newsroom in May, he argued AI must show it can improve Australians’ lives and strengthen the nation’s capability, not just cut costs. The bank says it spends about $2.4 billion a year on technology and capability, and has set aside a $90 million Future Workforce Program to reskill staff.
That framing has not settled the harder question. Comyn has acknowledged some roles will shrink as others grow, and CBA’s use of AI in its workforce has already drawn scrutiny. The bank drew public criticism in 2025 after moves to cut roles it linked to AI-driven efficiency, and the Finance Sector Union has pressed the major banks on how automation affects jobs and customer service. A rollout on this scale sharpens the tension between the productivity story and the employment one.
Why it matters for Australia
CBA is setting a template. As the country’s biggest bank and one of its largest private employers, its decision to treat generative AI as standard-issue tooling gives other large institutions a concrete reference point, and cover to move. Where earlier corporate AI news leaned on cautious trials, this is a whole-of-workforce commitment with fraud and customer-facing operations explicitly in scope.
It also raises the stakes on questions Australian regulators and workers are still working through: how AI decisions are governed in a heavily regulated sector, how customer data is protected inside a US-headquartered model provider, and how a bank reconciles reskilling promises with headcount reality. Financial services sits under the eye of APRA and ASIC, and a deployment this visible will shape expectations for governance and disclosure across the industry.
The near-term test is whether the fraud and scam-response work delivers measurable protection for customers, and whether “agent-powered” service means faster help or fewer people to provide it. CBA has said new AI-enabled services for retail and business customers are due in the months ahead. Those releases, and the bank’s next workforce updates, will show whether the largest AI rollout by an Australian institution becomes the sector’s blueprint or its cautionary tale.
Sources: OpenAI, OpenAI for Australia, Commonwealth Bank newsroom, Gend, ARN.
