AI Brief Australia
The Index · Release · Published 9 September 2026 Revised 9 September 2026

Australian AI Media Tone Index, August 2026

In August 2026, 44.4% of AI stories at the front door of Australia's major mastheads were framed negatively, against 30.9% across the global comparison set: Australian coverage ran 13.5 points more negative. Based on 421 Australian and 946 global stories.

The gap
+13.5 pts
Australian negative share minus global
Australia, negative
44.4%
n = 421, ± 4.7
Global, negative
30.9%
n = 946, ± 2.9

Shares

Australia, Tier 1n = 421
44.4% negative · 37.1% neutral · 18.5% positive · ± 4.7 on the negative share
Global setn = 946
30.9% negative · 39.7% neutral · 29.4% positive · ± 2.9 on the negative share

By topic

Topic · share negativeAustraliaGlobal
Jobs and work 56% 60%
Safety, ethics and regulation 69% 65%
Business and investment 25% 17%
Government and public sector 16% 12%
Education 62% 38%
Health 43% 9%
Crime, scams and deepfakes 95% 95%
Science and research 14% 0%
Products and consumer 30% 6%
Culture and media 55% 42%
Politics 14% 25%
Infrastructure and energy 38% 18%

By masthead

The share of each masthead’s AI stories framed negative and positive. These are aggregates about a masthead’s framing, not claims about any person. We show fewer than twenty stories as insufficient.

MastheadSetnNegativePositive
ABC NewsAustralia11551%8%
Australian Financial ReviewAustralia10932%28%
The AustralianAustralia7941%27%
news.com.auAustralia5147%16%
Sydney Morning HeraldAustralia4957%14%
The AgeAustralia18insufficientinsufficient
ReutersGlobal34225%38%
Wall Street JournalGlobal22329%30%
Financial TimesGlobal17129%28%
BBC NewsGlobal9445%22%
Washington PostGlobal9042%12%
New York TimesGlobal17insufficientinsufficient
The GuardianGlobal9insufficientinsufficient

Examples

Headlines as published, with links. We characterise framing, not people.

Positive framing

Cite this release

AI Brief Australia (2026). Australian AI Media Tone Index, August 2026. Australian AI Media Tone Index. https://aibrief.com.au/tone-index/2026-08. Published 9 September 2026.

In a sentence
In August 2026, 44.4% of AI stories at the front door of Australia's major mastheads were framed negatively, against 30.9% across the global comparison set: Australian coverage ran 13.5 points more negative. Based on 421 Australian and 946 global stories.

Data: CC BY 4.0. Method and rubric version on this page; the CSV carries every figure.

Method

Rubric 0.1 · prompt 0.1 · classifier claude-opus-5 · human-check agreement 98%
Rows by route: Google News search 99%, RSS 1% · 95% interval on the Australian negative share ± 4.7 points
We read the headline and standfirst only, never the body. Full method · Download CSV

First published release, from the back-run. Human check: fifty rows per set coded blind by the editor gave 72% (Australia) and 70% (global) agreement, with every disagreement on the neutral boundary and polarity crossing once in a hundred rows; the editor then recoded the disagreements strictly to the rubric, knowing which rows they were, and agreed with the model on all of them (98% and 100%). That second figure is an upper bound; from September the check is coded to the rubric blind from the start. Rows were collected by Google News RSS search after the fact and carry no standfirst, so this history is classified from headlines alone. The global set is carried by three business wires; the New York Times and the Guardian are under-collected in this version and the next version fixes that.

Revisions

  • 9 September 2026Eight August-dated stories arrived by the live RSS route on 9 September, after first publication; the release now includes them. Previously gap +12.4 pts, Australia 43.3% negative (n = 413), global 30.9% (n = 946).