Latest: Release says Sarvam's five-year defence MoU covers battlefield decision-making, target acquisition and surveillance data processing
HQ Integrated Defence Staff, under the Defence Ministry, signed a five-year MoU with Bengaluru-based AI firm Sarvam for AI projects across the Army, Navy and Air Force. Sarvam's Chanakya division will work on decision support, target acquisition, data fusion and surveillance data processing, with human oversight on operational decisions, according to the company and a release. Summary updated 28 Oct, 18:25 IST
Why it matters: The pact shows how India is bringing homegrown AI into military use, raising questions of national security, data sovereignty and oversight.
Winding down: no new coverage since 28 Oct. It will be archived on 4 Nov unless new coverage appears.
New developments are added here, newest first. When the headline, summary or other sections change, the earlier version is kept under “See earlier version”.
Headline before this update: Defence staff HQ signs AI pact with Sarvam for tri-services
Summary before this update: Sarvam said it signed an MoU with HQ Integrated Defence Staff, under the Defence Ministry, for AI projects across the Army, Navy and Air Force. Its Chanakya division will work on decision support, target acquisition and data fusion, with human oversight on operational decisions.
Note before this update: Coverage is consistent and relies mainly on the company's announcement, differing only in emphasis on sovereignty, funding and technical detail.
“What's missing” before this update:
Based on 4 articles read in full. Scores describe each article's framing, not the outlet and not whether it is accurate.
Large outlets read: 1 of 15 (Hindustan Times). Most of India’s largest outlets, including NDTV, India Today, Times of India, Hindustan Times, The Hindu and Indian Express, block our reader in this prototype, so their coverage is missing here. The launch pipeline reads their public feeds.
Coverage is consistent and relies mainly on the company's announcement and a release, differing only in emphasis on sovereignty, funding, battlefield use and technical detail.
No real disagreement between sides in this coverage.
Why: Frames the pact as part of Atmanirbharta and indigenisation, with national-security and self-reliance emphasis, relying only on the company's announcement; no critical voices or questions raised.
Why: Plain report of a defence-AI MoU, but relays sovereignty and Atmanirbhar self-reliance framing and national-security aims largely from the company and HQ IDS, with no critical voices.
Why: Wire report of a defence AI MoU, framed around Atmanirbharta, indigenisation and sovereign infrastructure. Leans right on security and self-reliance, but it is a plain announcement with only company sourcing.
Why: Plain report of a defence MoU relaying the release; sovereign, India-built AI and national security framing, and private-sector partnership lean mildly right, but there is no praise or criticism.
Critical, Neutral or Supportive, scored separately. It refers to the Union government unless the story is about a state. On the spectrum strip, each dot is an article: its position is the ideology score and its colour is the stance.
If an article would score differently with the parties' roles reversed, that signal is partisanship. It goes into the stance tag, not the ideology score.
High, Medium or Low, based on how much of the article was read, how many clear signals it has, and whether they agree.
Stories with fewer than five articles are marked. A blindspot is a contested story where we found coverage from only one side.
Articles that touch none of the three dimensions, such as a car launch or a match report, carry no ideology score.
This prototype is refreshed by hand (last on 17 Oct, 03:00 IST). At launch:
| Wires and the top 20 outlets | checked every 5 minutes |
| All other outlets | checked every 15 minutes |
| New story | goes up at 3 articles from 2 or more outlets, marked Limited coverage |
| Live stories | new articles added every 15 minutes; summary rewritten at most hourly |
| Home page order | every 15 minutes |
New facts go into a dated update log at the top of the story. New articles carry an “Added” tag. If a note or list changes, the earlier version stays viewable.
A story is Live while coverage arrives, Winding down after about 2 days without new coverage, and Archived after about 7. Nothing is deleted: links never change, archived scores are frozen, and stories stay on their issue pages and in search.
NDTV, India Today, Times of India, Hindustan Times, The Hindu, Indian Express, News18, ABP, Times Now and Mint block the reader used for this prototype, so their articles are missing. Each story shows how many of India’s 15 largest outlets we read. The launch pipeline will read outlets’ public feeds, TV transcripts and Hindi articles.
Editor-in-charge: S. Kalra, editor@thenazariya.com. The editor-in-charge sets Nazariya's editorial rules, reviews the AI desks' daily decisions, and is responsible for everything we publish.
How we work: AI reads and scores every article against our published rubric; an AI standards desk checks accuracy and fairness; people set the rules, review decisions and handle complaints.
Grievance officer under the IT Rules, 2021: to be confirmed (based in India). Until then, write to grievance@thenazariya.com. Complaints are acknowledged within 24 hours and resolved within 15 days. To report a factual error: corrections@thenazariya.com. Other questions: support@thenazariya.com.
Ownership: Nazariya is independently owned by S. Kalra, an Indian citizen.
Funding: Nazariya is currently funded entirely by its founder. It carries no advertising and takes no money from any political party, government or party-linked organisation. In future it will be supported by optional reader memberships. Any other funder, grant or investor will be listed here, with the amount and date.
Independence: No owner, funder or member can change a story, a score or a correction. Scores follow the published rubric, and anyone can appeal them.
Anyone, including the outlet that published an article, can appeal a score: tap any score on a story page, then "Appeal this score" to write to the editor. Outlets can also dispute a summary. The editor reviews every appeal, and any change is recorded below. Appeals: appeals@thenazariya.com.
No detail that could identify a child or a survivor of sexual violence (Juvenile Justice Act, POCSO, BNS section 72). Court cases are reported as what was said and alleged. No analysis of a poll in the 48 hours before voting there (Representation of the People Act, section 126). Press Council and NBDSA norms on communal incidents.