Claim reviewed
Our reading: Example City Police ran live facial recognition on people attending the anti-government march of 11 October 2026.
The article’s main takeaway is that live face matching was used on the marchers themselves. Its factual premises are that the force deployed the system for that march and that the cameras compared the faces of people in the procession, not merely passers-by in the same streets.
Support
The article links two documents the force published under its transparency policy. The deployment notice (fictional source S1) names the 11 October march as the purpose of the deployment, gives the operational window of 14:00–17:30 and lists three camera positions on the march route. The post-deployment audit log (fictional source S2) records that live matching ran for the whole window, that the cameras were directed at the march column, and that 4,120 faces of people in the procession were compared against the force’s watch-list.
Together the two documents establish the rated takeaway as the article states it: live facial recognition was run on the people attending that march, at that time. The article quotes both documents accurately and does not go beyond them in its headline or opening paragraphs.
What the evidence does not establish
The audit log records no identifications, stops or arrests, and the review makes no finding about any individual. Neither document says the deployment was lawful or unlawful, secret or announced beyond the notice itself, or whether the march was peaceful; the article does not claim those things in its rated takeaway and the badge does not rate them. A green result is a finding about evidence support, not approval of the surveillance.
Other findings and limits
The example covers the named takeaway and its two premises. The article’s later paragraphs quote campaigners’ reactions; those are opinions and were not rated. The source descriptions below are fictional summaries, not retrieved documents or verbatim extracts. No actual intake, source inspection or access attempt took place.
Review method and corrections
Illustrates an AI-assisted evidence review under Editorial Standard rules 1.2 using the locked report structure: a defined takeaway, the material support, what remains outside the finding, other findings and limits, and method and correction context. Example revision 1, dated 20 October 2026. No correction recorded.
Fictional source summaries
S1 — Example City Police deployment notice, operation EX-2026-11, published 10 October 2026. Scenario source description, not a quotation: states that live facial recognition will be deployed for the anti-government march of 11 October 2026 between 14:00 and 17:30, with cameras at three named positions on the published march route.
S2 — Post-deployment audit log, operation EX-2026-11, published 15 October 2026. Scenario source description, not a quotation: records the system running throughout the window, cameras directed at the march column, 4,120 faces of people in the procession compared against the watch-list, and no record of identifications, stops or arrests.
Demonstration only · No real article review.