Shield · Synthetic — NEW · 7th Shield · Founding customers Q3 2026

The deepfake is already inside
your video KYC queue.

Nearly half of Indian adults have already encountered an AI voice-clone scam attempt — roughly twice the global average per Phobolytics / ORF (2026). India digital-fraud losses crossed ₹22,931 cr in 2025 per the MoHA I4C portal — a 40× rise since 2021. RBI's V-CIP processes over 11 lakh video KYC calls every day, and synthetic-media injection is increasingly observed against this surface (CERT-In Sept-2025 advisory). Pindrop and Reality Defender lead the global market — neither speaks Hinglish.

Aria's Synthetic Shield is the only India-native answer: V-CIP injection detection, Hinglish red-team-as-a-service, and a synthetic exposure score that plugs into your Underwrite shield. Built by the only team in India that ships an attacker pipeline daily — so we know what these attacks actually look like in 2026.

Request founding-customer pilot See Red Team service
₹22,931 cr
India digital fraud losses, 2025 — up 40× since 2021
Ministry of Home Affairs · I4C
47%
of Indian adults already hit by an AI voice-clone scam (2× global avg)
Phobolytics · ORF, 2026
11 L+
RBI V-CIP video KYC calls processed daily — every one a potential injection target
RBI Master Direction
3 hr
IT Rules Feb-2026 mandatory deepfake takedown clock
MeitY
Three modules · one shield

A complete synthetic-defense stack — priced for India, billed in INR.

Each module ships independently and earns its keep — but together they form a feedback loop: the red team produces labelled attack data, the V-CIP gate detects it, the exposure score prices it for your insurer.

MODULE · 01
V-CIP Injection Detection
API · per-scan billing

A frame-level synthetic-media gate that sits in front of HyperVerge / IDfy / Signzy / in-house V-CIP flows. Catches injection attacks (DroidCam, virtual-camera bridges, replay) and generated faces before the video stream reaches your KYC verifier.

  • Real-time per-frame scoring · 60 ms p95
  • Injection vector detection (virtual-cam, replay, frame-stitch)
  • Generative-face detector ensemble (Hive + open-source backup)
  • Every flagged event auto-feeds CERT-In Annexure-II flow
  • Mumbai data residency · footage never leaves asia-south1
From ₹3 L / month per mid-size bank · scales with V-CIP volume
OEM/embed pricing for KYC platforms — contact sales
MODULE · 02
Synthetic Red Team-as-a-Service
Engagement · ₹15-40 L

A quarterly, board-ready synthetic attack engagement. We scrape your CEO/CFO public footage, generate Hinglish deepfake voice + video using our production HeyGen + voice-clone stack, and run multi-channel attacks against finance, treasury and IT helpdesk. Then we hand you the report, mapped to CERT-In categories.

  • Hindi / Marathi / Tamil / Bengali deepfake assets
  • Multi-channel attack: WhatsApp + email + voice call
  • Helpdesk social-engineering script library
  • 7-day SLA (industry standard: 21 days)
  • Every engagement → labelled attack data → trains Module 01
₹15 L Pilot · ₹25 L Standard · ₹40 L Enterprise
Quarterly retainer available · 4-engagement bundle ₹1.4 cr
MODULE · 03
Synthetic Exposure Score
Underwrite extension

Bolt-on to your existing Underwrite shield. Computes a synthetic-exposure sub-score (public exec footage volume, voiceprint availability, V-CIP coverage, Hindi/regional voice clone risk) and ships it as an HMAC-signed monthly delta to your cyber insurer. Module 03 launches once Module 01 has 12 months of in-tenant detection data — actuarial validity requires real loss correlation.

  • Exposure index v1 — claim-validated score in v2 (2027)
  • HMAC-signed monthly delta payload
  • Insurer-side dashboard (white-labelled)
  • Premium simulator: "what if we ran a quarterly red-team?"
  • Channel pricing: insurer pays, end-customer benefits
Bundled with Underwrite shield · channel revenue share
Insurance carrier partnership program — contact partnerships
Why Aria · why now

The moat isn't a model. It's the bundle.

No single capability wins this category. The attacker has voice + video + WhatsApp + Hindi + UPI fluency — your defence needs all of them under one roof. Here's why every other vendor is missing at least one piece:

CapabilityAria CyberShieldReality Defender / PindropHyperVerge / IDfyIndian MSSPs
Hinglish CEO deepfake factory✓ daily, in production
V-CIP injection detection✓ frame-level~ video-only~ liveness only
WhatsApp-native delivery✓ Meta Cloud API~
CERT-In Annexure-II auto-draft✓ 6h clock~ manual
asia-south1 data residency✓ Mumbai-only✗ US/EU
INR / GST / Razorpay billingUSD
Multi-agent SOC + ledger✓ 6 other shieldspoint toolpoint tool~ services
Indian-language attack corpus✓ Hindi/Marathi/Tamil/BengaliEnglish-only

The attacker pipeline

We run a faceless-AI content factory daily. Every avatar we ship for our own brand becomes training data for Module 01's detector and inventory for Module 02's red-team library. The content business funds the R&D of the security product.

The regulator pipeline

Every flagged synthetic event lands in the same evidence vault that drafts your CERT-In Annexure-II, your DPDP §8(6) breach notice, your IRDAI 48-hour clock. Detection without reporting is half a product — we're the only ones with both.

The distribution pipeline

HDFC ERGO / ICICI Lombard / Bajaj Allianz cyber-insurance products have to underwrite synthetic exposure by 2027 — they have no model for it today. Module 03 makes us the scoring engine they cite. Compounding moat: copying us means displacing the insurer relationship.

One Tier-2 bank. 90 days. ₹15 L pilot.

We're opening three founding-customer pilot slots for Q3 2026 — ideal fit is one Tier-2 bank or large NBFC with a real V-CIP surface, one mid-cap listed company with public exec footage, and one BFSI cyber-insurer. You get a 90-day co-designed pilot, the ₹15 L pilot fee credits in full toward your first annual contract, and (with mutual consent) you're cited as a reference logo for the broader market launch.

Apply for founding pilot Or start with a Red-Team engagement
3 founding pilot slots open 90-day co-design ₹15 L credits to annual contract NDA-first conversation