Australia’s largest bank is the latest employer to be embarrassed by its own automation plans. The Commonwealth Bank of Australia (CBA) is currently being forced to rehire staff after laying them off on the basis that their jobs had been made redundant by artificial intelligence. The about-turn comes after a challenge by the Finance Sector Union (FSU), which alleges the bank misled both its workers and the public about the effect of its new chatbot.

First, the bad news. The CBA reportedly approached dozens of workers in March to tell them that their roles were no longer needed. The bank’s reason? Its recently introduced AI “voice bot” had cut the number of incoming calls by about 2,000 per week, and that reduction was the main reason it no longer required as many workers. The layoffs targeted several long-serving workers at the bank, with some having worked at the organization for as long as 30 years.

But employees reacted with disbelief to the story, noting that it did not chime with the facts. The reality, they say, is that the volume of calls was not going down at the time but, in fact, going up. CBA management had been complaining to employees of the increased workload, even redeploying managers to work on calls and offering overtime to those who did.

The union referred the matter to a Fair Work Tribunal, which agreed to hear the case. The FSU argued that CBA had failed to properly explain to workers why their roles had been declared redundant. It also accused the bank of potentially using the announcement to obscure a separate decision to move the roles offshore to India, while simultaneously making new hires in that country. The suspicion was that the chatbot announcement was a red herring intended to distract people from the fact that the bank was now outsourcing these roles to India.

The bank later conceded a crucial point. In the tribunal, CBA admitted that its staff had missed the increased call volumes in making the case for redundancies. The volume of calls had, in fact, risen sharply, up by 2,000 a week when compared to the period before the chatbot was announced. This change was ongoing, having continued for months, and ran directly counter to the reason originally given by CBA for the redundancies. “This error meant the roles were not redundant,” the bank said in the tribunal.

Bowing to the inevitable, the bank has had little choice but to backtrack. CBA has since apologized to the workers involved and said that the 45 employees who were laid off will be given the option to return to their old roles, apply for new ones within the company, or take an exit package. “We have apologized to the employees concerned and acknowledge we should have been more thorough in our assessment of the roles required,” a CBA spokesperson told Bloomberg.

The FSU is hailing the decision as a “massive win” for its members. However, the union is also saying that the damage to workers has already been done. For weeks, the 45 workers at the center of the case were forced to wonder about the future of their employment, while others suddenly found themselves unable to pay their mortgages and utility bills. The union is saying that cases like this risk causing a lot of collateral damage to innocent workers as the rush to automate carelessly gathers pace.

The controversy is a small but growing storm in a much bigger shift across banking. Banks all over the world are rushing into using AI, which is helping to drive estimates that between 100,000 and 200,000 jobs could be cut in banks over the next three to five years, Bloomberg Intelligence estimates. The areas in which banks expect to make most cuts are back office, middle office, and operations, but banks are also said to be bracing for disruption in core areas like sales and trading.

Chatbots could prove an essential part of the cost savings, and CBA is not the only bank currently ramping up investment. Even as it publicly reversed course on the AI cuts, the CBA announced a new partnership with OpenAI just last week. That agreement will see the two companies work together on advanced generative AI tools to drive more robust scam detection and fraud prevention, as well as improve and personalize customer service. But despite CBA’s assurances that the move was designed to “invest in our people and embed the responsible use of AI,” employees will be skeptical after this latest fiasco.

The controversy comes as banks around the world grapple with large-scale change. The risk of backlash from employees being told their jobs are done by AI will surely be on many bankers’ minds as their employers race to implement the technology. “The bank has dodged a bullet on this occasion, but the case also has a very important message,” says FSU assistant national secretary Beth Niven. “AI technology is important, but employers need to be responsible about its use.”