Newsroom wire photo triage
Score wire photos, social-media-sourced images, and reader-submitted visuals before editors cite them in breaking-news copy. The cost of running a synthetic image as 'photo from the scene' is measured in retraction posts and Pulitzer board reviews.
Business value
- Adds a triage layer before the photo-desk decision, especially in breaking-news situations where editors don't have time to verify every reader-submitted image.
- Catches the synthetic-photo pattern that's specifically targeting newsrooms — generated 'eyewitness photos' uploaded to social during fast-moving events.
- Documents the verification step for editorial accountability. When a photo is published with provenance gaps, the gate's evidence array is part of the audit trail.
Agent job to be done
Be the photo desk's first reader on uncertain images. Allow wire photos with established provenance (AP, Reuters, AFP staffer credit). Hold reader-submitted and social-sourced images for source verification. Hold breaking-news images regardless of score until provenance is confirmed.
format: articleintended_use: citedomain: newsroom wire photo triage
Newsroom verification standards: AP, Reuters, AFP, and the Trust Project
AP's verification guidelines require multiple-source confirmation for any 'photo from the scene' caption. Reuters' Standards & Values explicitly cover synthetic-image risk and require staff to flag uncertain imagery. AFP runs a dedicated fact-check team. The Trust Project's indicator framework names 'image provenance' as a publishable trust signal. The image-scoring gate slots into all four — it doesn't replace the verification standards, but it makes the verification step measurable and auditable.
When to call VeracityAPI
On photo intake from any source other than a known wire credit. Always run on reader submissions, social-sourced grabs, and stringer uploads.
What image URL to submit
Public HTTPS image URL for the photo as it will appear in publication. If the photo will be cropped or color-graded, score the version closest to publication.
Decision policy
- allow: low risk AND verified wire credit OR confirmed stringer source.
- revise: medium risk OR reader-submission without established provenance — initiate source verification (request original file with metadata, contact submitter).
- human_review: high risk, OR breaking-news imagery (regardless of score, breaking news requires verification before publication), OR images that would caption as 'photo from the scene' or 'eyewitness footage.'
- reject: high risk combined with a source the newsroom can't verify, OR image is being used to support a specific factual claim about events.
Request template
The exact payload shape this use case sends. The sample below uses representative content for this workflow; substitute your own.
curl https://api.veracityapi.com/v1/analyze \
-H "Authorization: Bearer $VERACITY_API_KEY" \
-H "Content-Type: application/json" \
-d '{"type":"image","content":"https://veracityapi.com/demo/influencer-beauty-tonic.jpg","context":{"format":"article","intended_use":"cite","domain":"newsroom wire photo triage"},"store_content":false}'Automation recipe
- Photo desk receives the image and its purported source.
- Worker scores the image with intended_use=cite (which raises the threshold for medium-risk content).
- Allow-tagged images from verified wires flow to layout.
- Medium and high-risk images route to source verification with the evidence categories pinned for the verifying editor.
- Verification artifacts (source confirmation, original file metadata) get stored with the publication record.
Evidence spans agents should inspect
- 'synthetic_scene' — generated 'eyewitness' scenes with characteristic generation cues (overly painterly composition, impossible lighting in a 'photographed' moment)
- 'composite_seam' — figures pasted into scenes that don't agree on color cast or shadow direction
- 'metadata_strip' — image arrives without EXIF that would be present in actual camera capture
- 'social_recompression' — image has passed through multiple platform recompression cycles, which masks both authentic and synthetic cues — flag for additional verification
Policy pseudocode
if (result.recommended_action === "allow") continueWorkflow(); if (result.recommended_action === "revise") rewriteWith(result.evidence, result.recommended_fixes); if (result.recommended_action === "human_review") queueForHumanReview(result); if (result.recommended_action === "reject") discardOrRebuild();
KPIs to track
- % of non-wire photos held for verification
- verification-completion rate before publication deadline
- post-publication corrections involving image provenance (the metric this gate drives to zero)
- median time from photo intake to publication-ready
- editorial confidence on a periodic survey of photo-desk staff
What can go wrong
- The gate doesn't replace photo-desk judgment. It surfaces signals; editors decide.
- Phone-camera EXIF is increasingly stripped by social platforms. 'Metadata strip' alone isn't proof of manipulation — it's a prompt to verify by other means.
- Wire credits aren't perfect. AP and Reuters have published synthetic images in the past after being deceived by stringers. The gate is useful even on wire imagery for high-stakes stories.
Cost and latency notes
Image analysis is a flat $0.02 per image. The endpoint accepts HTTPS image URLs, stores no image bytes, and logs only a URL hash plus hostname. Current v0.1 latency is vision-model-bound, so preflight balance and retry carefully.
Agent evaluation checklist
- Does this workflow have a costly failure mode from manipulated or synthetic-looking visuals?
- Can the agent store evidence categories without storing raw image bytes or full URLs?
- Should this workflow fail open, fail closed, or queue human review if image scoring is unavailable?
- Which field drives policy: recommended_action, risk_level, content_trust_score, or synthetic_image_risk?
- What local provenance check should complement the API score?