Whether you can AI screen offshore candidates properly depends on what the model is reading. Most tools rank CV language against a job description, which mainly predicts how well someone writes a CV. For a role where the written artefact is the work, that is a fair proxy. For nearly everything else it is a guess wearing a confidence score.
In December you will have to write down which of those decisions your software is making. That tends to concentrate the mind. Outstaffer sells an AI recruiter, so weigh what follows accordingly.
From 10 December your privacy policy names the screening
The Office of the Australian Information Commissioner confirms that the automated decision-making transparency obligation commences on 10 December 2026, introduced by the Privacy and Other Legislation Amendment Act 2024. Entities using personal information in automated decisions that could affect someone's rights or interests have to set out, in the privacy policy, the kinds of personal information used and the kinds of decisions made with it.
The Privacy Act covers organisations turning over more than $3 million a year. Most businesses reading this are inside that line. The ones under it are not.
Rejecting a candidate affects their interests. The vague answer about what the tool does stops being tenable in about twelve weeks, so if you are still working out what an AI recruiter does once the marketing copy is stripped away, now is the time.
What an AI screen actually reads on an offshore candidate
Take a thirty-person business hiring six roles a year. Four hundred applications a role at ninety seconds each is ten hours of reading, sixty hours across the year. A screening model hands almost all of that back. Price it at whatever your own hour is worth. That saving is real.
Then look at the other column.
Does AI screening reduce the cost of a wrong hire?
Only if it ranks on something that predicts performance. Research SEEK released in May 2026, conducted by advisory firm Nature across more than 950 small businesses in Australia and New Zealand, put the cost of one wrong hire at about $16,000 on an $80,000 salary, with wrong hires staying 11 to 14 months against roughly three years for the right ones. Outstaffer has not verified that dataset independently, so read it as reported rather than official.
Get one in four wrong across six roles and the year costs about $24,000. The faster shortlist did not touch a dollar of it.
Is AI resume screening biased?
It inherits whatever the training signal carried. A model ranking CV phrasing favours candidates who write like the people already hired, which is a real problem inside one market and a sharper one across several. A Manila bookkeeper with eight years of Xero work writes nothing like a Sydney one, and the model reads the difference as a gap in capability.
The honest case for AI screening
It is strong and worth putting properly. Volume triage is what a model is good at, ranking 400 down to 40 in minutes, the same way every time. A tired human at application 380 is measurably worse than the same human at application 12.
The argument was never that the screen is worthless. It is that the screen is the cheap half, and the all-in monthly cost of an offshore employee shows how much is riding on the other half.
Related reading
Judge the tool on retention, not on speed
Ask any vendor two questions. What signal does the model rank on, and what happened to the people hired through it twelve months later. Answering the first and not the second means you are buying a sorting algorithm. Privacy obligations turn on your own circumstances, so get advice on yours before December.
My position is that AI belongs at the top of the funnel and a person belongs at the bottom, and that any claim to screen offshore candidates properly without a human validating the shortlist is a claim about cost rather than quality. To test it, run one role through ours and judge the shortlist yourself.
One question worth arguing with. If your last three hires had been ranked by a model instead of read by you, would you have hired the same three?
More from
Global Hiring
category
%201.svg)
