Ping Identity research put the average threshold at which Australian consumers would hand an AI agent purchasing authority — without seeking final sign-off — at around AUD $200 for everyday transactions. More than half of those surveyed — 53 per cent — placed a ceiling of AUD $143 or below on what they were prepared to put toward that category of expenditure.

Data for the local findings were drawn from 1,000 Australian consumers who took part in a broader multinational survey covering 11,000 respondents spread across 12 countries. Australian data point to a reserved public stance toward AI once it moves from fielding queries into taking actions on a person's behalf.

Among those surveyed in Australia, just under one in five expressed openness to AI making decisions on their behalf without human input, and more than a third ruled out granting AI any degree of the responsibility levels put to them. One in four respondents worldwide declined to assign any level of responsibility when surveyed on the topic.

Transparency around AI use drew particularly strong backing from respondents in Australia, where 68 per cent said they wanted disclosure every time the technology was deployed — a proportion that outpaced every other nation included in the survey.

Alongside these findings, the study pointed to a broader pattern of doubt among respondents about how well organisations protect and manage personal identity information. Across the surveyed pool of Australian consumers, only one in fourteen expressed complete confidence in how organisations handle personal identity data.

Near two in five survey participants from Australia — 42 per cent — flagged AI as a threat to the security of their personal data, with a closely matched 43 per cent pointing to privacy intrusion as a separate concern. Regulatory protection of personal data in AI systems ranked as a priority for consumers, with four out of five Australians surveying as supportive of oversight measures in that space.

About two-thirds of those surveyed had little to no familiarity with what the term agentic AI actually refers to. Sharing access to core personal services was something many buyers proved unwilling to do. Fewer than one in four respondents indicated a willingness to grant an AI agent read-level access to their main inbox, a figure that held even when a five-hour weekly time saving was placed on the table.

Three in four Australians expressed unease at the prospect of an AI agent acting on their behalf in ways that might distort how their character or core values come across to others.

A survey finding that 43 per cent of respondents expressed greater unease about needing to constantly verify an AI agent's outputs than about the agent making outright errors — a concern cited by 31 per cent — was accompanied by a statement from Darryl Jones, who holds a consumer segment strategy role at Ping Identity, noting that while people appear open to expanded AI capabilities, they want input into where boundaries are drawn, and that as AI takes on broader responsibilities, transparency around its permissions, data access, and lines of human oversight becomes increasingly important.

Across the world, more than half of those surveyed said they felt at ease turning to AI for everyday assistance or product guidance. Willingness to act, however, fell away once any practical steps were on the table. Across the global survey pool, just one in twenty respondents expressed comfort with AI acting after receiving approval, and that share fell to roughly one in thirty for routine transactions or actions carried out without any final human sign-off.

Survey data placed respondents in India at 70 per cent and those in Indonesia at 61 per cent among those willing to let AI act autonomously on their behalf. Rates of 21 per cent in Sweden, 19 per cent in the Netherlands, and 18 per cent in Australia were recorded across those three countries.

Demand for transparency about AI use rated highly among consumers in Australia, the United States, and the United Kingdom, whereas a survey finding from Indonesia pointed in a different direction — six in ten respondents there placed greater weight on speed and accuracy than on disclosure.

Respondents spanning every surveyed market reported that 71 per cent harboured unease about AI agents distorting their personality or values when acting on their behalf, and nearly half — 47 per cent — identified the ongoing need to monitor an agent's output as a greater source of concern than any error the agent itself might produce. Across all respondents surveyed globally, the share expressing complete confidence in organisations' handling of personal identity data fell to 12 per cent, a drop of five percentage points from the prior year's figure.

The survey underpinning the findings was carried out by Talker Research on behalf of Ping Identity, drawing on a double-opt-in methodology with random participant selection. Respondents drawn from twelve countries — the US, UK, France, Germany, Australia, Singapore, India, Indonesia, the Netherlands, Sweden, Spain, and Italy — formed the consumer base for the research.