CBA AI job losses warning pushes banks into workforce planning

Matt Comyn is telling Commonwealth Bank staff that generative AI will leave some teams smaller, a message Australian executives usually keep buried in consultant decks. In an Australian Financial Review opinion piece, the CBA chief executive said large employers should stop offering false comfort about what the technology will do to white-collar work. He is not talking from the fringe: Comyn runs the country’s biggest bank, oversees heavy technology spending and has already tied AI to workforce design, retraining budgets and productivity targets.

Capital Brief reported that he also told staff there was “no use pretending otherwise” if the likely outcome in some areas was smaller teams. For workers, that sounds less like experimentation and more like a sign the AI debate has moved from pilots and copilots to payroll. The unease is easy to follow after ABC reported that CBA cut 45 roles during an AI voice-bot rollout, then reversed course after call volumes rose.

Taken together, the message lands as an enterprise story rather than a culture-war one. A blue-chip Australian employer is putting headcount, retraining and copyright in the same frame. That matters more than another generic claim that AI will change everything.

From pilots to payroll

In the AFR piece, Comyn argued that employers of CBA’s scale should prepare staff for change rather than pretend the old shape of work will hold. That places the bank in a different camp from companies still treating generative AI as a bounded experiment inside support teams or coding groups.

He wrote in the Australian Financial Review that “pretending otherwise does not protect workers. It only ensures they are surprised later”.

By February, CBA said it would spend $90 million over three years on a Future Workforce Program, with more than 30,000 employees receiving AI-focused training. The same announcement said 5,000 staff had already moved into new roles inside the bank over the previous year. Those numbers suggest a company planning for disruption rather than denying it.

From an analyst’s perspective, the real question is whether AI lifts productivity enough to justify redesigning roles around it. CBA has not published a simple ratio between automation gains and headcount cuts. It has published a budget, a training plan and an internal mobility number. The transition is already being measured in operational terms.

Earlier this month, an AFR Technology analysis made a similar point across employers including CBA, PwC, BHP, Telstra and Freehills: the market has started treating AI as a return-on-investment project, not a science experiment. Comyn’s latest warning pushes that logic further. Once management talks about returns, staff will ask what happens to the roles attached to the costs.

The worker case is harder than the boardroom case

Boardrooms can tell a tidy story about orderly adjustment: train people early, move them into adjacent jobs and use AI to lift output per employee. Inside large service operations, workforce change rarely feels that neat. One automation gain can create work somewhere else.

Recent history at CBA explains the scepticism. The ABC’s reporting on last year’s backtrack showed why early automation stories can mislead executives as much as workers. The bank cut roles after introducing an AI voice bot, then discovered that higher call volumes and operational pressure complicated the case for a clean reduction. AI can remove tasks, but it does not always remove demand.

Union concerns also sit behind the tension. Finance Sector Union national secretary Julia Angrisano has previously warned that staff want clarity on job security, not just broad promises about innovation and adaptation. Comyn’s February workforce announcement framed the programme around “transparency and opportunity”, and in March he said in a Commonwealth Bank article that “Ultimately, this technology needs to lead to productivity and improvements in living standards”. Those lines show management knows trust has become part of the implementation problem.

That defence only works if the gains show up in service quality, wages, new roles or some combination of the three. If they do not, other employers can still borrow the cost-cutting language and skip the expensive retraining programme CBA is at least trying to build.

Copyright has become part of the labour story

Copyright is the other revealing part of Comyn’s intervention. In the AFR’s reporting on his intervention, he urged Australia to protect copyright as it tries to attract frontier AI investment. That shows how quickly workplace AI has become entangled with the policy bargain Canberra is being asked to strike.

Canberra wants access to major model builders and the economic activity that comes with them. Reuters reported in March that Anthropic was signing a deal with Australia on AI safety and economic data tracking. ABC News later reported on the government’s push to lure the company deeper into the local market. The same discussion also carries demands around copyright, data access and regulatory certainty.

That leaves a narrow path. Give too much away on copyright and Australia weakens one of the few levers local creators and publishers still have. Give too little and policymakers risk hearing that the country is closed to investment just as banks, law firms and corporate IT teams begin reorganising work around the new tools. Comyn’s intervention matters because it joins those two tracks. The labour consequences are no longer separate from the rules around the models.

For digitalblog’s enterprise lens, the takeaway is straightforward. CBA is showing how Australian boardrooms are likely to talk about AI in the next phase: fewer abstractions, more workforce planning, and a stronger insistence that regulation should help large employers deploy the technology at scale. Once that script is public at a bank this large, it will not stay a banking story for long.