UnbiaslyAI
AI-verified news for India

Read the news.
See the source.
Decide for yourself.

India's first AI-powered news platform with source attribution and framing analysis on every story. Built for 900 million internet users who want fast information they can trust.

Try the analyzer→Read the live feed
Live · Last analyzed story
Source verified
The Hindu
VerifiedEN
Headline · Framing analyzed

Parliament passes amendment after extended debate; opposition stages walkout citing procedural concerns.

The bill passed 287 to 142 after four hours of debate. Three opposition parties left the chamber before the vote, calling the process rushed. Government called the bill essential for the next fiscal quarter.

Framing flag
Loaded framing

The highlighted phrase carries an interpretive load. We surface it so readers can weigh the framing rather than absorb it.

Word count · 60
Cycles every six seconds · Three real stories from the pipeline
900M+
Indian internet users
10K
Stories processed daily
60
Words per summary
4.5★
Android beta rating
Built and advised by
Carnegie Mellon University
IIT Delhi
LSE · University of London
Government of India · 16 years
MIT · IBM · Meta
Stanford · CodeX
India Today Group
AI4Bharat embeddings
Sarvam APIs
Indus AppStore launch partner
Carnegie Mellon University
IIT Delhi
LSE · University of London
Government of India · 16 years
MIT · IBM · Meta
Stanford · CodeX
India Today Group
AI4Bharat embeddings
Sarvam APIs
Indus AppStore launch partner
Try it yourself

Paste any headline.
See the framing.

This is a small slice of the live pipeline. It scans for loaded language, missing attribution, and sensational framing. The full system also runs source verification, claim tracing, and native Indic-language analysis.

Try a sample
Framing analyzer · demo
Input
Annotated reading

Government slams critics as opposition desperately scrambles to defend its embarrassing record on inflation.

Framing score
72
/ 100
Attribution
Needs source

Heavy framing. Multiple loaded signals present.

Flags · 4
  • slams · Emotionally loaded verb
  • desperately · Imputed motive
  • scrambles · Imputed motive
  • embarrassing · Editorial judgment
Demo runs locally in your browser · Full pipeline also analyzes Hindi and Bangla
About UnbiaslyAI

An app built to combat misinformation and rebuild trust in what India reads.

UnbiaslyAI curates reliable content from authentic national and international sources. Every story is delivered as an easy to understand 60-word summary with full source verification. Our pipeline reads each story in its original Indian language wherever possible. This allows bias that lives in idioms, honorifics, and local framing to be detected before it reaches the reader.

We are recommendation first, not publisher first. The feed learns what you read, skip, share, listen to, watch, and return to. It then recommends source attributed content in your preferred language and format. Over the next twelve months, the same verified stories will be available as text, AI generated audio, and AI anchor short form video. Verified journalists and creators will publish through the same provenance layer and earn through a revenue share model as trusted inventory scales.

Download the app
Free. Available on iOS, Android, and Indus AppStore. Hindi, English, and Bangla.
Download on the App StoreGet it on Google PlayGet it on Indus AppStore
Inside the UnbiaslyAI app
How it works

The pipeline behind every story.

Most misinformation in India does not start as completely fake news. It starts as selective framing, emotional phrasing, missing context, or sourceless forwarding. Our pipeline runs four stages of work between a publisher hitting publish and a reader seeing the story.

01

Ingest

We pull stories from authentic Indian and international sources. The pipeline processes around 10,000 articles a day across Hindi, English, and Bangla.

02

Analyse in language

Each story is read in its original Indic language wherever possible. Bias lives in idioms, honorifics, and local framing, all of which are lost when content is translated to English first.

03

Attribute and flag

Every claim is tied back to its source. Loaded language, one sided framing, missing attribution, and unsupported claims are flagged before content reaches the feed.

04

Summarise

Each story is condensed into a 60-word summary with the source visible. Readers can scan the summary, click through to verify, and decide for themselves.

15 to 30 min
Publisher to feed
99.5%
Crash-free sessions
<0.2%
Summary error rate
22
Indian languages
Today on UnbiaslyAI

The pipeline is running right now.

Live · last update 0s ago
Stories analyzed today
9,456
Live, updating every few seconds
Framing flags raised
247
Across all three languages
Languages processed
3
Hindi, English, Bangla · 19 more on roadmap
Verified sources active
480
National and regional publishers
Where we sit

Not another news app. A trust layer for the ones you already use.

Indians already get news from many places. The problem is not access. The problem is provenance, framing, and language. Here is how UnbiaslyAI compares with the platforms most readers use today.

Capability
UnbiaslyAI
Inshorts
Dailyhunt
Google News
X / Twitter
Source attribution on every story
Framing and bias analysis
Native Indic-language processing
We read in Hindi and Bangla without translating to English first
No paid placements in the feed
Recommendation-first feed
60-word summaries with source
Verified creator program
Yes Partial No
Why now

The trust layer is missing and the gap is widening.

India crossed 900 million internet users while the supply of unverified, AI-generated, and emotionally framed content scaled past every existing safeguard. The window to build a native solution is now, not later.

01
78%
Of Indians get news from WhatsApp forwards

Closed group sharing strips out source attribution. By the time a story reaches the reader, the origin is gone.

02
5x
Growth in AI-generated news content since 2023

Synthetic articles, AI-narrated videos, and fabricated quotes now flood Indian-language feeds. Detection lags creation.

03
22
Official languages, almost no native AI trust tooling

Most fact-checking and framing tools assume English. Bias that lives in Hindi or Bangla idioms is invisible to them.

The wedge

Indian attention does not live in “news.”

It lives in six high frequency categories that existing apps treat as side categories. We treat them as the main wedge. This is where Indians spend time every day, and it is also where framing and rumours spread fastest.

A
ज्योतिष
Astrology

Festival timing updates and life-event guidance, with named source astrologers.

B
बॉलीवुड
Bollywood

Film, music, and OTT coverage with paid-placement detection on every review.

C
क्रिकेट
Cricket

Live scores, IPL coverage, and player news pulled from verified beat reporters.

D
भक्ति
Devotion

Temple events and sacred calendars, attributed to recognised religious bodies.

E
अर्थशास्त्र
Economics

Markets, policy, and personal finance with source pills on every claim.

P
राजनीति
Politics

National, state, and constituency-level coverage with framing flags applied.

“Native Indic neutralisation is a moat, not a feature. Bias in Hindi or Bangla cannot be detected by translating to English first. It lives in idioms, honorifics, and local framing.”

UnbiaslyAI Engineering
Features

Six things that make a feed worth trusting.

We do not claim to determine the truth. We show what was reported, who reported it, and how it was framed. Readers can then form a better judgment.

60

60-word summaries

Every story is condensed into a single scannable paragraph with the source clearly visible. Readers cover more news in a minute than in an hour of scrolling.

✓

Source attribution

Every article links back to a verified publisher. We surface who reported it, when it was reported, and how. Stories are never anonymous and never sourceless.

↔

Framing analysis

Loaded language, one sided framing, and missing attribution are flagged in real time. Making bias visible is the first step toward addressing it.

अ

Indic-first AI

The platform supports Hindi, Bangla, and English today, with nineteen additional languages on the roadmap. The system is built on AI4Bharat embeddings, with a proprietary Indic news SLM in development.

◯

Recommendation first

The feed is recommendation first rather than publisher first. It learns from what each user reads, skips, shares, listens to, watches, and returns to.

Coming
▶

Audio and video

AI generated audio and AI anchor short form video for verified stories will roll out across Indian languages over the next twelve months.

Built by

The people behind the pipeline.

Two ends of the same problem. Umesh from Carnegie Mellon on the model side. Ashwani with 16 years inside the Government of India on the distribution side. Eighteen months of building together, always in person.

01
Umesh Sharma

Umesh Sharma

Co-Founder and CEO
Carnegie Mellon University

Built the original AI pipeline. He owns model architecture, the neutralisation evaluation system, and the proprietary Indic SLM roadmap.

Profile→
02
Ashwani JP Singh

Ashwani JP Singh

Co-Founder and COO
16 years, Government of India

Owns strategy, growth, institutional distribution, partnerships, and editorial operations. He is also the founder of Green Governance India.

Profile→
03
Saharsh Sikaria

Saharsh Sikaria

Chief Technology Officer
IIT Delhi

Owns the application stack, infrastructure, ingestion systems, and ML serving. He works directly with Umesh on the AI integration layer.

Profile→
04
Kriti Agarwal

Kriti Agarwal

Co-Founder and Chief Content Officer
LSE, University of London

Owns content strategy, editorial standards, and the verified creator program. She sets the bar for what gets published and how.

Profile→
05
Robin Singh

Robin Singh

Co-Founder and Chief Business Officer
University of Delhi

Owns revenue, partnerships, and go to market. He leads B2B verification pilots, institutional channels, and the monetisation roadmap.

Profile→
Honorary advisors

Counsel from journalism, governance, and AI policy.

Uday Mahurkar

Uday Mahurkar

Former Information Commissioner, Government of India
Former Deputy Editor, India Today Group
Anil Rao, IPS (Retd.)

Anil Rao, IPS (Retd.)

Former CID Chief, Haryana
Vice President, DAV College Managing Committee
Riyanka Choudhury

Riyanka Choudhury

AI Governance at MIT, IBM, and Meta
Fellow, CodeX, Stanford Center for Legal Informatics
The supply side

A platform for verified journalists, creators, and institutions.

Trust scales when supply does. Verified regional creators, journalists, and domain experts will publish through the same provenance layer that powers our news feed. They will earn through a revenue share model as trusted inventory grows.

01 / JournalistsComing

Trusted reporters, trusted reach.

Verified regional journalists publish directly through our provenance layer. Their stories carry source attribution, framing analysis, and native language coverage from day one.

02 / CreatorsComing

Domain experts, monetised.

We are onboarding verified creators across the ABCDEP categories. Creators earn through a revenue share model as trusted inventory scales, while provenance is preserved.

03 / InstitutionsComing

Public information, properly distributed.

Government departments, public information channels, and civic bodies can reach citizens through verified channels in their native language.

Are you a journalist or creator?
Early access applications are open. We will be in touch.
Apply for early access→
What is next

Trust scales in formats.

Text was the start. Over the next twelve months we extend verified content into audio, video, and conversational interfaces. Provenance is preserved across every format.

Format 1Live
Text

60-word summaries.

Available now across Hindi, English, and Bangla. Every story scannable with full source attribution.

Format 212 months
Audio

AI generated audio.

Verified stories listenable in Indian languages, for commutes, kitchens, and accessibility. Powered by Sarvam APIs.

Format 312 months
Video

AI anchor short form video.

Watch the same trusted summary in thirty seconds. The provenance layer stays the same, only the format changes.

Get the app

See the source. Always.

Free to use. Available on iOS, Android, and the Indus AppStore. Read more than 10,000 stories a day with full source attribution and framing analysis.

Download on the App StoreGet it on Google PlayGet it on Indus AppStore
orRead the live feed →