Mainstream media coverage of artificial intelligence in Australia has grown from a trickle to a torrent in the past year. But while reporting over that period makes clear the business case for AI adoption and attracting data centres, the benefits to workers and consumers has not filtered through.
And for all the talk about building local AI capability, the most influential voices on AI and data centres have come from offshore, and less than 1 per cent of coverage involved an Australian-made AI.
With the release of new research tracking media coverage and sentiment over the past 12 months, you get the sense that Aussies look at AI as technology that is being done to Australia, not for Australia or by Australians.
The report, prepared by iSentia in partnership with InnovationAus.com, covered more than 230,000 pieces of news coverage between August 2025 and August 2026. You can read the full report here.
It found that over the course of the year, the coverage became steadily more confident that AI technology works, but steadily more doubtful that the benefits of that technology would benefit ordinary people.
The research compared that coverage with more than 13,000 posts from Australian subreddits, which provided an indicative picture of online public discussion rather than a representative opinion poll.
Key news moments over the 12-month period included the Treasurer’s Economic Reform Roundtable, the under-16 social media ban, the Grok deepfake harms and Prime Minister Albanese’s ‘AI in Australia’s Interests’ address at the University of Sydney in July.
Isentia executive director of AI strategy Ngaire Crawford says the data paints a clear picture: When media coverage named a beneficiary of AI, it named business in 71 per cent of cases compared with naming consumers as the beneficiary in just 30 per cent, and workers in 26 per cent.
“The central unresolved tension is not whether AI works, but whether its benefits are shared fairly. The costs – including job cuts, higher household bills and environmental pressure – are reported as falling on workers, consumers and communities, often without those affected having a voice or any solution being offered,” Ms Crawford said.
“This divide is already sharper in public discussion on social media. Jobs and work accounted for 30 per cent of AI discussion in the Australian online communities analysed, compared with just 5 per cent of news coverage.
“We expect these concerns to move increasingly from social channels into mainstream news over the next 12 months.”
The divergence was also evident across the report’s four elements of trust. Among news coverage that weighed each issue, 53 per cent was supportive of the governance response and 20 per cent critical.
By contrast, 80 per cent was critical of the technology itself and 19 per cent supportive. The public discussion analysed was more critical across all four elements.
Government was the largest source of headline and opening comment on AI, accounting for 21.4 per cent.
The Prime Minister’s speech in July at the University of Sydney was also the only government initiative to cut through to the sampled public discussion, with the busiest week of online AI discussion coinciding with the peak in news coverage of the plan.
InnovationAus.com publisher Corrie McLeod said the research showed the government’s voice strengthening as it developed regulation, safeguards and new institutions, but the wider debate remained heavily dependent on international voices.
“Government became more visible as it addressed regulation, safeguards and the organisations needed to oversee AI, particularly through the National AI Plan [speech] in July. But beyond government, the voices shaping the debate are largely international,” Ms McLeod said.
“Two-thirds of major stories were international news relayed to Australian audiences, and the vast majority of quoted AI industry voices were global. Australian-built AI companies accounted for less than 1 per cent of coverage.
“Despite all the discussion about sovereign AI, the media conversation gives very little attention to the local capability Australia will need if it is to build and shape this technology for itself.”
The research also found that about one in five stories concerned an AI-related harm or failure, but 63 per cent named no responsible actor. Where responsibility was assigned in a news story, it fell most often on individual users and businesses deploying AI, and least often on the developers that built the systems.
The coverage therefore often left no clear indication of who was accountable for rectifying the harm.
Coverage of data centres grew roughly tenfold across the year. By July 2026, stories focusing on costs to individuals – energy bills, water use and grid strain – outnumbered those focusing on benefits to individuals, including jobs, by about four to one. The companies building the centres held only a small share of voice.
Key data points
- Two-thirds (66 per cent) of major stories were international news relayed to Australian audiences; 88% of quoted AI-industry voices were international; and Australian-built AI companies appeared in 0.8 per cent of coverage
- When coverage named an AI beneficiary, it named business in 71 per cent of claims, consumers in 30 per cent and workers in 26 per cent (articles could name more than one)
- Jobs and work accounted for 30 per cent of online AI discussion, six times their 5 per cent share of news coverage
- About one in five stories concerned an AI harm or failure, and 63 per cent of that coverage named no responsible actor
- Government supplied 21.4 per cent of headline and opening quotes, the largest share. Prime Minister Albanese was the most-quoted individual and the person most consistently framed positively
- Data centre coverage grew roughly tenfold; by July 2026, cost-to-individuals framing was about four times as prevalent as benefit-to-individuals framing
Do you know more? Contact James Riley via Email.