Why did Government gloss over use of AI against beneficiaries?

Their proclamations that they always had the power to use AI as a weapon against beneficiaries is concerning…

Ministry says it could always use automated decision-making to make welfare decisions

The Ministry for Social Development says it has never been restricted in its use of automated decision-making in the welfare system, after a bill providing a broad permission to use it passed through Parliament on Friday.

– Advertisement –

The amendment to the Social Security Act, which was passed under urgency as part of the Government’s 2026 Budget measures, expanded the permitted uses of automated decision-making (ADM) to any social assistance programmes, and any statutory powers, duties or functions exercised by the Ministry.

Previously, provisions of the Social Security Act allowed the use of ADM for “specified provisions” if they were about one of four things:

1. Child support payment info shared with the Ministry (and how that affects a person’s assets or income).
2. Breaches of obligations.
3. Mandatory reviews.
4. The expiry and re-grant of certain benefits.

But after questions from Stuff, the Ministry has revealed they believe they were always allowed to use ADM more broadly.

Stuff

…as is their ramming it through with as little scrutiny as possible…

Cabinet agreed on AI benefit reforms a year ago, waited until Budget to pass under urgency

The Green Party is questioning why the Government rushed consequential reforms of the benefit system into law without a select committee during a few days of Budget urgency when Cabinet agreed to reform the law more than a year ago.

The Greens want to know why the Government legislated the changes under urgency last week despite having a year to introduce them after they were first announced in Budget 2025 last May.

The Government has admitted that passing the bill through all stages under urgency will allow it to meet its forecast savings target from the scheme, which are meant to be $225.8 million over four years, beginning this year. A full parliamentary process could have delayed this by about six months.

NZME

…I think this has happened because if Kiwis understood how MSD is using AI as a means to find reasons to cut people off welfare regardless of their circumstances they would turn on the Government, not because most Kiwis don’y enjoy kicking beneficiaries when they are down, but because Kiwis fear the Government using AI against them.

The Government changed the rules around welfare, forcing beneficiaries to reapply every six months rather than annually, and because the agencies don’t have enough staff to process those applications, they are now allowing automated systems to do it.

Beneficiaries are always the lab rats for how far the State thinks it can bend human rights, they are already under a level of mass surveillance we wouldn’t agree with if used on us.

They are guinea pigs in this experiment and that experiment is already working for the Government by slashing $55 million from the poorest…

$55 million less in benefit payments as Govt automates welfare decisions under urgency

It comes as new figures in the Budget reveal the Government expects to pay out $55 million less in benefits over the next four years as a result of the automation drive.

Green Party social development spokesperson Ricardo Menéndez March said the legislation would make it harder for people to access benefits and get paid the amounts they need.

Newsroom

…nothing comforts a struggling family like a chatbot saying ‘Have you tried not being poor?’

“Nicola Willis thinks AI can handle mental health support — because when you’re at your lowest, nothing hits quite like a machine asking you to rate your despair out of five.

National’s dream is a fully automated public service: no public, no service, and best of all, no one left to explain why it’s not working.

Automated decision-making doesn’t replace bureaucracy. It replaces accountability.