An evidence report compiled in support of Aletheia — AI Forensic Analysis Engine
Smart India Hackathon 2026 | Problem Statement ID: SIH260001
Team X — Mrinal Roy · Swastika Shaw · Rahul Sah · Somasree Giri · Arpita Kirity · Arjun kahar
August 2026
THE DEEPFAKE CRISIS | Team X | SIH 2026
Why This Report Exists
Numbers alone don't move judges, policymakers, or users — but numbers plus real people do. This report pulls
together the scale of the deepfake problem (growth data, financial losses, detection-accuracy research) and the
human cost behind it: named, dated, sourced cases of real people who lost money, dignity, or safety to synthetic
media, plus a digest of the news coverage documenting this crisis as it has unfolded through 2023–2026.
Every figure and story below is drawn from a cited, dated, publicly available source — no numbers or incidents
have been invented. Where a victim's real name was withheld by the original news outlet for privacy, that is
noted.
1. The Scale of the Problem
900%
growth in online deepfake volume, per year (WEF)
$2.7B
lost to cyber fraud in India in 2025 (I4C)
92,000+
"digital arrest" scam cases in India (2025)
47%
of Indian adults hit by an AI voice/deepfake scam
The volume of synthetic media online has been compounding at roughly 900% a year since 2019, according to the
World Economic Forum — a trend independent trackers confirm is continuing, with online deepfake volume
estimated to reach into the tens of millions by 2026.
Online Deepfake Volume Has Grown ~900% a Year
Deepfake videos/images online (millions)
0.014M
2019
0.500M
2023
8M
2025
14M
2026 (proj.)
Source: World Economic Forum (2023); The Global Statistics, Deepfake Statistics 2026; DeepStrike (2026)
Detection is not something the human eye can be relied on for. Across multiple independent 2025–26 studies —
University of Reading/Leeds/Greenwich/Lincoln, a Wiley-published replication, and public detection quizzes with
tens of thousands of participants — unaided human accuracy at spotting AI-generated faces sits in the low 30s
to mid 50s percent, essentially a coin flip.
THE DEEPFAKE CRISIS | Team X | SIH 2026
Unaided Human Judgement Is Barely Above Chance
Trained humans (super‑recognisers, after 5‑min training)
chance level (50%)
64%
Untrained humans (general public)
31%
020406080100
Accuracy identifying AI-generated faces (%)
Source: Gray et al., "Training human super-recognisers' detection and discrimination of AI-generated faces," Royal Society Open Science (2025).
2. Real People, Real Harm
These are not hypothetical risks. Each case below happened, was reported by name or verified detail in
mainstream or trade press, and represents a pattern repeating at scale across India and globally.
Ordinary Investor · India
"B. Hari", Bengaluru — Chartered Accountant
₹23.2 lakh lost
A Bengaluru-based chartered accountant saw deepfake videos of Mukesh Ambani, N.R. Narayana
Murthy, and Sudha Murty on Facebook promoting a stock-investment scheme in late 2024. The videos were
entirely AI-generated but looked and sounded authentic. He filled out a form, was contacted by a "financial
advisor" over WhatsApp video call from a UK number, and was walked through a series of transfers before
realising the platform and the executives endorsing it were both fake.
Source: Deccan Herald, "CA falls prey to deepfake video scam, loses Rs 23 lakh" (2025).
Digital Arrest · India
92-Year-Old Man, Delhi
₹2 crore drained
Fraudsters impersonating officials from the Telecom Regulatory Authority of India and
Maharashtra Police told a 92-year-old Delhi man that FIRs had been filed against him and he faced imminent
arrest. Kept under sustained psychological pressure, he transferred over ₹2 crore before Delhi Police's cyber
unit (IFSO) traced and recovered the funds — an outcome most victims do not get.
Source: Delhi Police / All India Radio News, "Delhi Police Busts Digital Arrest Scam, Recovers ₹2 Crore" (15 May 2025).
THE DEEPFAKE CRISIS | Team X | SIH 2026
Digital Arrest · India
Retired Merchant Navy Officer & 100-Year-Old Father, Lucknow
₹1.29 crore lost
A caller posing as a CBI officer told a 100-year-old man that a money-laundering warrant
had been issued in his name, then kept him under a six-day "digital arrest" over WhatsApp video call. His
70-year-old son, fearing his father's arrest, transferred ₹1.29 crore across multiple transactions before the
family realised it was fraud.
Source: Deccan Herald, "Ex-merchant navy officer, 100-year-old father lose Rs 1.29 crore in digital arrest scam" (28 Aug 2025).
Family-Emergency Voice Clone · USA
Jennifer DeStefano, Arizona
$1 million ransom demanded
Jennifer DeStefano received a call in which she heard what she was certain was her
daughter's voice, crying and saying she had been kidnapped, followed by a man demanding a $1 million ransom.
Her daughter was safe the entire time — her voice had been cloned from social media clips. DeStefano later
testified before the U.S. Senate about the incident, helping drive national attention to AI voice-cloning
fraud.
A finance employee at Arup's Hong Kong office joined a video call with people who looked
and sounded exactly like the company's UK-based CFO and several colleagues. Every participant on the call was
an AI-generated deepfake built from executives' publicly available conference footage. The employee
authorised 15 wire transfers before discovering the fraud through an independent follow-up call to head
office.
Source: CNN Business (16 May 2024); CFO Dive (17 May 2024).
THE DEEPFAKE CRISIS | Team X | SIH 2026
Non-Consensual Imagery → Federal Law · USA
Advocacy That Changed U.S. Law
In 2023, AI-generated non-consensual explicit images of more than 30 female students were
created and circulated at a New Jersey high school. One student's family spent over a year publicly
advocating for legal reform after finding no clear path to accountability. Their advocacy directly
contributed to the federal "TAKE IT DOWN Act," signed into law in May 2025, which criminalises the
publication of non-consensual sexual imagery, including AI-generated deepfakes, and mandates fast platform
takedowns.
Source: CNN (Nov 2023); The New York Times (Apr 2024); CBS News (May 2025).
Political / Celebrity Misuse · India
Rashmika Mandanna Deepfake
A deepfake face-swap video placing the Bollywood actor's face onto another woman's body
went viral in November 2023, viewed by millions before being debunked. It became India's first mainstream
deepfake controversy and a direct catalyst for the government's subsequent IT Rules amendments on synthetic
content.
Source: Multiple mainstream Indian outlets (Nov 2023); Cybersecify (2026).
3. News Coverage Digest
A selection of the dated news coverage this report draws on — for further reading and independent
verification.
British engineering giant Arup revealed as $25 million deepfake scam victim
CNN Business · 16 May 2024
Confirms Arup as the victim of a Hong Kong deepfake video-call scam, with the company
stating that fake voices and images were used to authorise fraudulent transfers.
Teen Girls Confront an Epidemic of Deepfake Nudes in Schools
The New York Times · 8 April 2024
Reports on the Westfield High School case and the broader pattern of AI-generated
non-consensual imagery in American schools, and the slow institutional response.
New Jersey teen victimized by AI porn attends 'Take It Down Act' signing at White House
CBS News · May 2025
Covers the signing of the federal law criminalising non-consensual AI-generated sexual
imagery, tracing it back to the Westfield High School advocacy campaign.
CA falls prey to deepfake video scam, loses Rs 23 lakh
Deccan Herald · 2025
Details how a Bengaluru chartered accountant was defrauded through deepfake videos of
well-known Indian business figures endorsing a fake investment platform.
₹13 crore lost in 'digital arrest' scam: Delhi woman duped via fake police, judge on WhatsApp
Storyboard18 (via WSJ reporting) · 1 April 2026
A 77-year-old retired pediatrician was coerced over days of video calls by fraudsters
impersonating police and a judge, losing over ₹13 crore — illustrating the scale digital-arrest fraud has
reached in India.
India Scams 2026: Fake Cops, Deepfake Tycoons, and the ₹22,500-Crore Fraud Epidemic
ScamWatchHQ · 6 June 2026
A comprehensive data roundup of India's 2025 cybercrime losses, digital-arrest scam
patterns, and the deepfake-driven investment fraud impersonating high-profile business figures.
How can we combat the worrying rise in deepfake content?
World Economic Forum · May 2023
The widely cited source for the finding that deepfake content online grew roughly 900%
between 2019 and 2020, and the trend's implications for businesses and trust online.
Deepfakes and Financial Cybercrime: India's Multi-Layered Response
Observer Research Foundation (ORF) · 15 January 2026
Analyses India's regulatory and financial-sector response to deepfake-enabled fraud,
citing survey data that 47% of Indian adults have been affected by AI voice-cloning or deepfake scams.
Online deepfake volume has grown roughly 900% year-on-year since 2019 (World Economic
Forum, 2023).
Cumulative global deepfake-related financial losses reached an estimated $1.56 billion
through the end of 2025, with over $1 billion of that in 2025 alone (Surfshark analysis of AI Incident
Database / Resemble AI data).
The FBI's IC3 unit logged 22,364 AI-tagged fraud complaints for 2025, totalling roughly
$893 million in reported losses — investment fraud alone accounted for $632 million of that.
96% of deepfake pornographic videos identified in a 2019 academic study were
non-consensual, with the overwhelming majority targeting women (cited in the 2025 International AI Safety
Report).
Unaided human accuracy at identifying AI-generated faces sits between roughly 31% and 54%
across multiple 2025–26 studies — near chance level.
India
Indians lost an estimated ₹22,495 crore (~US$2.7 billion) to cyber fraud in 2025, with
complaint volume up 24% year-on-year to about 2.81 million cases (I4C / Ministry of Home Affairs data).
More than 92,000 "digital arrest" scam cases were reported in India through 2025, with the
Supreme Court estimating nationwide losses near ₹3,000 crore for the year.
47% of Indian adults have been a victim of, or personally know a victim of, an AI
voice-cloning or deepfake scam — nearly double the 25% global average; 83% of Indian AI-voice-scam victims
suffered a monetary loss.
India's IT Rules amendment (effective 15 November 2025) requires platforms to label and act on synthetic
content; rules effective February 2026 mandate takedown of flagged deepfake content within three hours of a
court or government order.
In 2025, Indian authorities deactivated 1.2 million SIM cards, froze 1.33 million mule
accounts linked to fraud, and recovered ₹5,489 crore.