We use AI tools selectively to assist reporting — never to replace human editorial judgement. Every AI-assisted output is reviewed and approved by a human editor before publication.
#Principles
- Human accountability: a named editor is accountable for every published piece, AI-assisted or not.
- Transparency: we disclose when AI was used materially in research, transcription, or generation.
- Verification: AI output is treated as draft material to be fact-checked, not as a source.
#Permitted AI use
- Transcription of recorded interviews (always reviewed against audio).
- Translation drafts (always reviewed by a fluent editor).
- Search + research assistance (results verified against primary sources).
- Headline + structure suggestions (final choice always human).
- Image enhancement (sharpening, denoising) — never content alteration.
#Prohibited AI use
- Generating fabricated quotes, sources, or events.
- Creating photorealistic images of real people without explicit illustration disclosure.
- Synthesising voices for audio without consent.
- Auto-publishing without a human editor approving every word.
#Disclosure to readers
Articles where AI played a material role (translation, transcription, image generation, structured-data summarisation) carry an "AI-assisted" disclosure label at the bottom describing what AI did and what the editor reviewed.
