AI job losses 2026: evidence trails CEO layoff claims

An Apollo economist says there is still no broad evidence that artificial intelligence is destroying white-collar jobs at economy-wide scale. The claim matters because AI has moved from productivity pitch to redundancy rationale, giving executives a cleaner story for restructures that may have been coming anyway.

For Australian technology workers, the sharper read is not that AI is harmless. It is that the public language around layoffs is running ahead of the labour data: fewer roles, flatter teams, more automation, and an AI explanation attached to the memo.

Meanwhile, the people building the tools are retreating from the most severe forecasts. OpenAI chief executive Sam Altman told an audience in Sydney that he did not expect a jobs apocalypse, while Nvidia chief executive Jensen Huang has argued that blaming AI for cuts can be a lazy way to describe more complicated decisions.

Torsten Sløk, chief economist at Apollo Global Management, put the economist’s version of that pushback more bluntly in Business Insider’s report.

“zero evidence of job losses because of AI”
— Torsten Sløk, Apollo Global Management, via Business Insider

The evidence gap matters

Sløk’s point is narrow, but useful. The US labour market is not showing the kind of broad collapse in white-collar employment that would be expected if generative AI had already become a mass substitution technology. Business Insider noted that private payrolls rose by 110,000 in April, even as technology companies and office-heavy employers continued to attach AI language to job cuts.

A company can cut jobs in the same year it deploys AI. It does not follow that AI caused the cut. Interest rates, slower hiring after the pandemic boom, currency pressure, investor demands and management simplification can all sit behind the same announcement.

Apollo’s argument is also not a denial that tasks are changing. It is a warning against converting every restructure into evidence of a labour-market regime shift. The early signal is uneven: some roles are being compressed, some teams are being asked to produce more with the same headcount, and some firms are using AI to justify reductions that were already financially attractive.

Timing is the tell. Executives are getting the reputational benefit of sounding decisive about automation before the data can show whether the productivity gains are durable. Workers get the fear immediately. The proof arrives later, if it arrives at all.

Layoff memos are doing two jobs

Recent AI-layoff language is often doing two jobs at once. It tells investors that management is acting on the technology shift. It also gives employees a reason for cuts that sounds structural rather than discretionary.

Wix, for example, is cutting 20 per cent of its workforce, and its memo mentions AI alongside other pressures. Telstra is cutting more than 100 technology roles as part of a broader overhaul under chief executive Vicki Brady, according to The Australian. Those are real headcount decisions. They are not, by themselves, proof that AI systems have replaced the workers one-for-one.

Rhetoric is becoming risky because companies still need staff to use the tools. A Business Insider analysis cited survey work suggesting 99 per cent of C-suite leaders expect AI to translate into some job cuts over the next two years. If that is the message workers hear first, the adoption programme starts with suspicion.

Huang’s objection goes to that management problem. In Fast Company’s account, the Nvidia chief executive criticised the habit of treating AI as a simple explanation for reductions.

“the narrative that connects AI to job loss … is just too lazy”
— Jensen Huang, Nvidia, via Fast Company

Self-interest is part of the line. Nvidia sells the chips powering the AI buildout. Even so, it exposes the contradiction in the current corporate pitch: AI is described as a tool every employee must adopt, while the same technology is presented as the reason some of their colleagues are gone.

Australia is not outside the narrative

Australia’s technology labour market is smaller, but the same language is landing here. Telstra’s cuts show how quickly AI can become part of a local restructuring story, especially at large employers trying to simplify systems, reduce duplication and lift productivity.

As a case study, Telstra strips away some of the Silicon Valley noise. It is not a venture-backed AI start-up trying to impress investors with an automation story. It is a large incumbent with legacy systems, regulated infrastructure, enterprise customers and pressure to keep costs under control. AI may well help automate parts of that business. It is also a convenient shorthand for a wider operating model change.

For workers, the distinction is not academic. If a role disappears because a process is automated, retraining needs to be targeted at the new workflow. If it disappears because management is simplifying a division, AI upskilling alone will not solve the problem. A vague AI explanation can blur those two cases and make the adjustment harder.

Reuters framed Altman’s Sydney comments in that context. The OpenAI chief executive said he did not think AI would produce the kind of mass job destruction often forecast by its most worried critics.

“I don’t think we’re going to have the kind of jobs apocalypse”
— Sam Altman, OpenAI, via Reuters

No promise for every software engineer, support worker or analyst sits inside that quote. It is a narrower claim that the labour market may adapt through task changes, new demand and new forms of work, rather than through a single wave of replacement.

Productivity is the harder story

Productivity may be the more durable AI story, not immediate headcount destruction. The New York Times’ reporting on Schneider Electric described a manufacturer using AI in production work without turning the case study into a simple layoff narrative. The company has about 160,000 employees globally, and the example points to process change rather than a clean workers-out, software-in equation.

Higher output is not automatically benign. If a team can do more work with the same headcount, hiring can slow. Entry-level roles can thin out. Contractors can be cut first. Those effects may not appear as a dramatic unemployment spike, but they can still change career paths and bargaining power.

Apollo’s Jevons-style argument is that cheaper and faster work can create more demand. If software teams ship more features, marketing teams test more campaigns or support teams resolve more cases, the total volume of work may grow. That has happened with past technologies. It is not guaranteed to happen evenly, or quickly enough for people caught in the transition.

This is the gap executives should be forced to explain. Not whether AI exists in the workflow. It does. Not whether it can reduce some tasks. It can. The unanswered question is whether a specific round of layoffs follows from measurable automation, or whether AI has become the boardroom’s most fashionable cover story.

The management risk

Corporate rhetoric may backfire before the technology does. Companies want employees to test copilots, rebuild workflows and share internal data with new systems. That requires trust. Telling workers that AI is both their new assistant and the reason for the latest cuts weakens that trust.

Australian employers have a cleaner option: specificity. Say which tasks have been automated. Say which jobs are genuinely no longer needed because the process changed. Say which cuts are about cost, duplication, strategy or foreign exchange. If AI is only one input, do not make it the headline.

Evidence may change. Generative AI systems are improving, and some occupations will face deeper disruption as agents move from drafting text to executing work across business software. But in mid-2026, the stronger reading is that AI has become a management narrative faster than it has become a proven economy-wide job killer.

Workers and executives should both worry about that, for different reasons. Workers deserve a clearer account of why roles are disappearing. Executives need those same workers to believe the next AI deployment is a tool, not just the preface to another redundancy round.