For most of this year, the finance press has covered artificial intelligence in banking as a capability story: which bank shipped an agent, which vendor raised a round, which exchange added a model. Last week, the story that traveled across three continents of coverage was not about what AI can do inside a bank. It was about who gets retrained when it does the job instead.
The shift: a training pledge with a number and a deadline
On September 24, Singapore’s Deputy Prime Minister and Monetary Authority of Singapore chairman Gan Kim Yong announced the IBF AI Workforce Co-Lab at the Institute of Banking and Finance Distinction Evening. Twenty-three financial institutions, spanning banking, insurance and asset management and together employing more than 70% of the sector’s workforce, committed to train their entire Singapore headcount, more than 80,000 employees, in critical AI skills by 2028. More than half of that workforce has already completed IBF-recognised training. The Co-Lab runs on three tracks: Co-Lab@Leaders, which equips executives to identify AI use cases and set governance; Co-Lab@Wealth Managers, which upskills relationship managers at ten participating institutions; and Co-Lab@Operations, which redesigns back-office and processing roles at major banks and insurers as AI takes over repetitive work.
“The 23 pioneer institutions have committed to train their entire workforce in Singapore, totalling over 80,000 employees, in critical AI skills by 2028,” Gan said in his keynote address. He added that the goal is for “AI to translate into better careers for Singaporeans and a more productive workforce,” and that reaching it “requires commitment and action by financial institutions, as well as partnership and support from the government and unions.”
Advertisement
Why operations is the track to watch
Of the three tracks, Co-Lab@Operations carries the most weight, because it is the only one aimed at roles AI is already capable of doing end to end rather than roles AI merely assists. Co-Lab@Leaders is a governance and awareness track for executives who are not themselves being automated. Co-Lab@Wealth Managers upskills relationship managers, ten employer institutions deep, in a function that still depends on client trust an AI system cannot yet replicate on its own. Operations is different: document processing, reconciliation, transaction monitoring and back-office claims handling are precisely the high-volume, rules-bound work that agentic tools already perform at scale elsewhere in finance. Putting a named, measurable retraining track behind that specific category of role is Singapore’s regulator making an implicit admission that the other two tracks do not require: that some jobs inside member institutions will not exist in their current form by 2028, and that the state, not just the employer, has a stake in what happens to the people who held them.
Three outlets, three different stories
What makes this worth a second look is not the announcement itself but how differently the outlets that picked it up chose to frame it once it left Singapore. Insurance Journal, running Bloomberg’s wire account, led with risk. Its coverage is explicit that “the technology could replace jobs faster than workers can be retrained,” and frames the entire Co-Lab as a hedge against that possibility rather than a straightforward upskilling initiative. It also supplies context the announcement itself did not volunteer: finance accounts for roughly 14% of Singapore’s economy and employs about 200,000 professionals, and this program expands a narrower pledge from a year earlier in which three local banks alone committed to train 35,000 employees annually.
Xinhua’s account and the wire copy that followed it at outlets including PressNewsAgency took the opposite framing almost entirely. Neither uses the word displacement or names job losses as a live possibility anywhere in its coverage. Instead both lead with the tripartite structure, government, employer and union working together, and both list the specific functions the Co-Lab targets: customer service, financial advisory, credit underwriting, fraud detection, market analysis and software development. The emphasis is coordinated capability-building, not risk mitigation. Where Bloomberg’s account reads as a story about a threat being managed, Xinhua’s reads as a story about a policy success being executed.
Neither framing is wrong. Both are drawing on the same speech and the same numbers. What the split reveals is that even a government program with a hard number attached, 80,000 workers, a 2028 deadline, 70% of the workforce, does not settle the underlying argument about whether large-scale AI adoption in financial services is primarily an opportunity or primarily a threat to the people currently doing the work it will automate. The press covering it imported that argument rather than resolving it, each outlet reading the same commitment through the lens it already held.
What it means for the finance leader
The more useful comparison for a finance executive outside Singapore is not Bloomberg versus Xinhua. It is Singapore versus the regulatory language most other jurisdictions have used so far. FinTech wrote in an earlier opinion piece that the Financial Stability Board’s AI warnings carry no enforcement mechanism and function mainly as advisory language regulators can point to without requiring anyone to act on it. The Co-Lab is a different instrument. It is not a warning about AI risk in the abstract; it is a named commitment, with a headcount, a timeline and named institutions attached to specific tracks, that a regulator can audit against in 2028. That structure, a measurable workforce commitment rather than a risk disclosure requirement, is likely to be the more exportable template. Regulators in other markets watching their own banks deploy agentic tools, from AI agents that hold their own bank accounts to AI systems embedded directly in bank operations, now have a concrete precedent for pairing that deployment with a workforce obligation rather than leaving retraining to individual employers.
It also changes what “AI governance” means as a hiring and vendor question. A bank evaluating an AI deployment in a market that has adopted a Singapore-style commitment needs to ask not just whether the model is accurate and compliant, but whether the institution has a documented plan for the employees whose roles the model touches. That is a materially different due-diligence question than the one most compliance teams are currently asking, which still centers on model risk rather than workforce risk. Singapore is not the first place to raise agentic AI’s operational footprint in finance, and it will not be the last: a live charter race for AI-agent banking is already under way in other markets, and Singapore’s own fintech sector has spent the past year building out digital-asset infrastructure, including stablecoin rails that will themselves eventually need the same kind of workforce planning this program is trying to get ahead of.
The unresolved question
Neither the MAS announcement nor the coverage that followed answers the question that actually determines whether the Co-Lab succeeds: what happens in 2029, after the training is done, if AI has eliminated more roles than it redesigned. A training pledge measures inputs, people enrolled, courses completed, not outcomes, jobs preserved or created. Gan’s own language, that AI should translate into “better careers,” is a stated goal, not a guaranteed one, and none of the three accounts pressed the institutions making the pledge on how they will measure whether it worked. That is the gap a genuinely useful follow-up story, from any outlet, would need to close: not whether banks trained their staff, but whether the staff who went through Co-Lab@Operations still have a comparable job in three years. Until someone tracks that number, the 80,000 figure measures effort, not outcome, and the disagreement between Bloomberg’s risk framing and Xinhua’s opportunity framing will keep resurfacing every time a similar program launches somewhere else.
Source: Monetary Authority of Singapore
