సంపాదకీయ విధానం

Editorial policy

Sourcing and attribution

• Every story is based on a report that has already been published — by a news outlet, a broadcaster or another public source — or on copy filed by our own team.
• A story based on someone else’s report names that source and links the original (“Source: … · read the original report”). A story our own team wrote says it comes from the newsroom desk.
• The AI desk is instructed to use only facts that are in the source, and never to add names, numbers, dates or quotes. The Telugu, English and Hindi versions of a story must carry the same facts. An AI can still get things wrong, which is why every story links its source and why we correct mistakes.
• Photos and videos shown with a story come from its source’s report. Photos carry the source’s name, and the story links back to the report.
• Headlines are kept under 12 words, without clickbait or capitals for effect.

The desk check

• For every story, the AI desk lists two to five checkable claims from the source and notes how each one is supported — for example “attributed to the district collector”, “figure appears in the source” or “no source given”. Readers can open this list on the story (“What the desk checked”).
• It rates the story’s caution level: low for routine, attributed reporting; medium where significant claims are unattributed; high for sensational, unsourced or possibly misleading reports.
• The desk check works from the source alone, for attribution and consistency. It does not confirm claims with other sources or people. That is why a low-caution story’s label reads Desk check · compared with the source, never “verified”.
• High-caution stories are never published automatically. An editor decides, and must record a reason to publish one.

How we use AI

What the AI does

• Writes each story’s headline and summary from its source, in Telugu, English and Hindi.
• Runs the desk check, and tags each story’s topic and place.
• Writes the deep dive: background, key facts, a timeline and an exam angle, from the source only. Where the source does not say something, the brief says so rather than guess.
• Writes short analysis from a named angle — political, economic, legal and others. This is opinion, grounded in the story; it is labelled “AI” (for example AI Political Analyst) and is never presented as a real person’s words.
• Reads stories aloud in synthetic voices (Amazon Polly, and AI4Bharat’s Indic Parler-TTS for Telugu and other Indian languages).
• Makes some stories into short videos narrated by a synthetic voice. They are marked as AI: “AI-presented” where an illustrated AI anchor presents the story, “AI voice” on the others.

The writing is done by a large language model, Anthropic’s Claude.

What people do

• Choose the sources the newsroom follows and the rules it publishes by.
• Decide on held stories, and edit, hold, unpublish or delete any story at any time.
• Approve stories filed by our reporters before they are published.
• Write their own analysis. Our human analysts are named, and labelled “analyst”.
• Handle corrections and complaints.

What the labels on a story mean

Written by AI from <source> · checked by the newsroom

An editor published, approved or scheduled this story themselves in our newsroom system.

Written by AI from <source> · not yet reviewed by an editor

The story was published automatically — or released in bulk by the one-click button that publishes every held world story at once, which does not count as reviewing them, because nobody read them one by one. An editor may still correct or remove it later.

Stories do not carry a person’s byline: the text is written by the AI desk, and the label says whether an editor reviewed it. We do not invent human-sounding bylines, and AI analysts are always labelled AI.

Ethics

• Accuracy first. Only facts the source reports, the same facts in every language, and no invented quotes — the rules the AI desk is given, and the ones our editors hold it to.
• Opinion is labelled. Analysis sits apart from the reported facts and is marked as opinion. AI analysts are told to stay balanced, avoid partisan or inflammatory language, never state predictions as certainties, and say what the story does not establish.
• No words in real people’s mouths. An analysis may be written in a particular style, but it is never presented as something a real person said or wrote.
• No lottery or horoscope items on auto-pilot. Lottery, satta/matka, horoscope and astrology items are never published automatically.
• Mistakes are corrected openly, as our corrections policy describes.

Independence

Ooru News decides what it covers. Sponsors and advertisers do not choose, review or change our news stories. Paid content appears only as a labelled sponsored story.

Sponsored content

• A sponsored story is labelled “Sponsored” in the feed, and on its page “Paid placement by <sponsor>, labelled as such — not editorial.”
• Sponsored stories are never sent as alerts.
• A sponsor sees only the total views and clicks on its own story, never who read it.

Alerts

Alerts go only to readers who turn them on.

• Automatic breaking-news alerts. When a story that three or more sources carried is published, readers who turned alerts on get one. A story is alerted automatically at most once, there are at most three automatic alerts an hour, and the newsroom can switch them off.
• Editors’ alerts. An editor can also send an alert by hand for any published story that is not sponsored.
• A sponsored story is never sent as an alert, automatically or by hand.

Corrections and complaints

See our corrections policy, or contact the newsroom.