The headline evidence is mixed by design. This article reports what the available data can establish, then examines what it cannot see.

The AI Layoff Paradox: Australia Says “Not Yet” While Workers Lose Jobs

There are two Australian AI employment stories, and both are real.

The first comes from the federal government. The Department of Employment and Workplace Relations examined labour-market data through February 2026 and found no broad AI-driven upheaval. Overall conditions remained strong. Occupational reshuffling had not accelerated. Software-development employment had grown since November 2022.

The second comes from people receiving redundancy notices.

Atlassian announced 1,600 global job cuts in March 2026, including 480 Australian roles, and said the savings would fund AI and enterprise sales. WiseTech Global cut around 2,000 roles. Block announced thousands of global cuts, including a reported Australian impact. Telstra reduced roles in an AI joint venture with Accenture. The details and causes differ, but the worker experience is not an abstract statistical possibility.

The contradiction is not a reason to pick one side. It is the story.

A labour market can be healthy while a career feels broken

National employment statistics answer a broad question: are more people employed than before, and how are occupations changing overall?

An IT professional asks narrower questions:

• Can a graduate get a first job?
• Is a support role still a path into systems administration?
• Does five years of experience lead to progression, or only more work?
• Will a local team still exist after the next restructure?
• Does learning the company’s systems create security, or make someone easier to replace?

A national employment number can’t answer those questions by itself.

The government report found that employment in the most AI-exposed occupations grew by 5.6 percent between November 2022 and February 2026, compared with 9.5 percent in the least-exposed occupations. It estimated exposed jobs were about 2 percent below what a pre-ChatGPT trend would have predicted, while warning that the evidence was not definitive and that weaker growth predated ChatGPT.

That is an important warning, not a clean verdict. It suggests pressure may be appearing first as slower growth, fewer openings, and a more difficult transition rather than as an immediate collapse in total employment.

Why “AI caused the redundancy” is often too simple

A company may cite AI when several forces are operating at once:

• Investors want a faster path to margin improvement.
• A product is under competitive or market pressure.
• A previous acquisition increased headcount or complexity.
• Management over-hired during a boom.
• Customers are delaying projects.
• Work can be performed more cheaply in another location.
• A new AI capability creates a genuine reduction in certain tasks.

Those forces can overlap. It is possible for AI to reduce the amount of work required and for executives to use that reduction as an opportunity to make a much larger strategic cut.

That distinction matters because it changes the remedy. If the problem is only technological substitution, workers need transition support and new skills. If the problem is also financial extraction, workers need voice, bargaining power, consultation, and a share of the productivity gain.

Atlassian: a clear example of the new language

Atlassian’s announcement provides unusually direct public evidence. ABC reported that the company would cut 10 percent of its global workforce, including 480 Australian roles. The company said freed-up funds would be invested in AI and enterprise sales.

That doesn’t mean the 480 people were individually replaced by a model. It means the company made a workforce allocation decision in which AI investment and headcount reduction were connected.

The difference is more than semantics. “AI replaced my role” suggests a machine performed the work. “The company reduced the workforce to fund AI and pursue a new commercial strategy” describes the management choice more accurately.

Workers need accurate language because vague language weakens accountability.

The missing measure: who loses the first rung?

The most important effect may not appear in a redundancy total. It may appear in the entry-level pipeline.

Junior analysts, graduate developers, support technicians, testers, and coordinators often perform work that experienced staff have outgrown. That work is where people learn systems, judgement, client communication, incident handling, and the limits of theory.

If AI removes those tasks and companies remove the junior roles with them, the industry creates a strange future: fewer people receive the experience required to become the senior professionals that employers say they need.

The industry can cover the gap for a while by hiring experienced workers, outsourcing, or asking existing staff to do more. Eventually, the missing training pipeline becomes a capability problem.

A company can optimise its own headcount while weakening the profession it depends on.

Australia’s exposure includes offshoring

Australian workers are not facing automation in isolation. They are facing a global delivery model that has treated location as a cost variable for decades.

The Guardian reported that companies such as WiseTech, Block, and Atlassian were reducing staff while using AI to make remaining workers more efficient. It also reported that Telstra cut roles in its AI joint venture with Accenture, while the company said AI had not directly taken those roles.

That wording matters. “Not directly taken by AI” does not mean the work is secure. A role can disappear because management combines automation, process redesign, and offshore delivery. The worker experiences one outcome even when the executive presentation lists three causes.

A local IT workforce also carries knowledge that is difficult to count: customer relationships, regulatory context, informal troubleshooting, and the ability to notice when a technically correct answer will fail in the real environment. Remove too much of that capability and the company may save labour cost while increasing operational fragility.

What the government should measure next

The July report is a useful baseline. It should not be the end of the measurement programme.

Australia needs public reporting on:

• Entry-level hiring in AI-exposed occupations
• Graduate and apprenticeship conversion rates
• Redundancies and redeployment after AI deployment
• Wage movement by occupation and experience level
• Job vacancies, not only total employment
• Work intensity and hours for remaining employees
• The use of algorithmic performance and surveillance tools
• The local and offshore distribution of technical work
• Whether productivity gains are shared through pay, hours, training, or staffing

Without those measures, the country will discover a career crisis only after the people who would have filled the next generation of roles have already left.

What workers should do with contradictory evidence

Don’t let a reassuring national statistic invalidate what you’re seeing at work. Don’t let a viral layoff count persuade you that every technical career is over either.

Use the contradiction as a prompt to investigate your specific position:

1. Identify which tasks in your role are being automated, accelerated, or moved elsewhere.
2. Separate the tasks from the accountability. A tool may produce an output while you remain responsible for its errors.
3. Ask whether your employer has a redeployment and training plan, or only a list of efficiency targets.
4. Build evidence of outcomes that automation does not capture: risk prevented, incidents resolved, clients retained, systems made reliable, and decisions improved.
5. Keep your employment records, performance evidence, and redundancy documents in a private file.
6. Get advice before signing a separation agreement or resigning in response to pressure.

The government may be right that Australia has not reached mass AI unemployment. A worker may still be right that their profession has become less secure.

Both facts can sit in the same room.

Sources and method

Prepared August 8, 2026, using:

• The AI and employment in Australia report — DEWR
• Atlassian cuts 1,600 jobs — ABC News
• Will AI take Australian jobs, or is it just an excuse for corporate restructure? — The Guardian
• Cyber security skills in demand — Jobs and Skills Australia

Reported job-cut figures are attributed to the cited reporting. There is no official dataset that identifies the cause of every Australian redundancy as AI.

This is industry commentary, not employment or legal advice.

Related reading

• AI Didn’t Break Australian IT. The Growth Model Did.
• The IT Career Ladder Is Being Pulled Up Behind Us
• Reclaiming Work: A Survival and Power Plan for Australian IT Professionals