Editorial Standards
Hamidun News aggregates and adapts AI/tech news with AI assistance under human editorial oversight. These standards explain how.
01
Sourcing
- We aggregate from 42+ RSS feeds (TechCrunch, The Verge, Bloomberg Tech, Anthropic Blog, OpenAI Blog and others) and 8 first-party SPA sites.
- Every article cites the source via a prominent link and via structured data (the
isBasedOnschema.org property). - We never publish without naming the original source.
02
AI assistance
- Original article text is rewritten with Claude Haiku to a consistent house style (250–400 words for the short version, 800–1,500 words for the extended one).
- A quality-control layer filters transliteration artifacts (e.g. Russian «забут» for “booed”) and demotes low-quality rewrites (quality_score ≤ 4).
- Translations to en/es/pt/fr/ar use Gemini 2.5 Flash with empty-body detection and retry.
- AI does not invent facts; rewrites preserve the numbers, names and quotes of the source.
03
Editorial oversight
- Editor-in-chief: Zhemal Khamidun.
- Curation: a pyramid top-8-per-24h algorithm filters by quality_score; the Telegram channel receives the daily curated top.
- The morning digest is human-reviewable in the admin dashboard before broadcast.
04
Speed of correction
- Spotted an error? Email [email protected].
- We aim to acknowledge within 24 hours and fix within 48 hours.
- All corrections are logged on the Corrections page with a permanent entry.
05
AI ethics & transparency
- AI assistance is disclosed on every article via a badge under the byline.
- We follow the EU AI Act Article 50 transparency requirements for our AR/FR/ES audiences and apply the same rules to all languages.
- We do not use AI to fabricate quotes, sources or facts.
06
Contact
- Editorial: [email protected]
- Corrections: [email protected]
- Tips: [email protected] (or Telegram @JHamidun)
- LinkedIn: linkedin.com/in/jhamidun