It was billed as the next big step in financial technology, but for 45 staff at Australia’s biggest bank, it initially spelled trouble. The Commonwealth Bank of Australia (CBA) announced that its roles had been made redundant by a new artificial intelligence system. The workers, some of whom had been with the bank for decades, were soon told they would have to find new jobs or face the redundancy package. However, following a successful challenge by a union, the bank is being forced to reverse its decision.

The controversy centers on a new customer “voice bot” introduced by the bank to deal with customer calls. CBA claims that the chatbot has cut incoming calls by some 2,000 a week, and that reduced demand meant it no longer needed as many human employees. The decision to wind back headcount was reportedly welcomed by CBA, which says it will free up staff to focus on new tasks. However, the people whose roles were cut have less cause for celebration.

At a preliminary tribunal hearing, CBA appeared to concur with both claims, saying that its reasoning for redundancy had not considered an ongoing rise in call volumes. At the time of the layoffs, it turned out, call numbers were rising, not falling. The tribunal has heard that in some cases staff had been told to assist with answering calls, while others had been offered overtime. “This error meant the roles were not redundant,” CBA admitted in the tribunal. As a result of the admission, the bank has agreed to offer the affected employees a range of options. These include the opportunity to return to their old roles or other positions in the company, as well as redundancy payments.

In a statement, the FSU called the move “a massive win” for the staff affected, but warned that inaction had already taken its toll on the workforce. In many cases, it said, the bank had created weeks of uncertainty about paychecks, during which time some had not known whether they could pay their mortgages or meet other bills. In its challenge, the union said, it had intended to send a wider message to banks and financial institutions about the effect that sudden technological change can have on staff.

For now, the AI project appears to be continuing apace. Last week, CBA announced a new partnership with OpenAI to build tools to help prevent scams and fight fraud. “Our partnership with OpenAI will help us to build next-generation generative AI tools that will enhance our security systems, helping to detect scams and stop fraud faster,” the bank said. “This will also allow us to offer our customers more personalized and helpful services.” CBA added that the partnership was “about embedding the responsible use of AI and how this can benefit employees,” but did not comment on whether the case would affect its approach to AI.

This experience, however, comes at a time when banks in other jurisdictions are also planning to use AI and other automation measures to slim down their operations. A recent report by Bloomberg Intelligence suggests that as many as 200,000 financial services jobs could be cut worldwide over the next three to five years as automation takes hold. Banks are attracted to the prospect of the technology making their operations more efficient and cost-effective, but the CBA case shows that there is still a risk of reputational damage if plans are not fully thought through. The impact on trust, in particular, appears to be lasting. In a statement, the FSU said that “the damage has already been done,” and that many staff are likely to refuse the offer of returning to their previous roles.

In the meantime, however, the bank is facing another challenge from the union. It is currently appealing the terms under which CBA has consulted with staff on its plans to use AI. The terms on which the bank is using AI and has informed staff are at the center of the case.