The Real Buyers Behind a $15B Market
Market analysts classify Deepfake Detection as a cybersecurity sub-category. But its revenue structure and buyer personas look closer to BFSI KYC infrastructure than to security. It’s a category you can view through two different lenses on the same market.
This distinction matters operationally. Approached as cybersecurity, you target CISO (Chief Information Security Officer) budgets and security-team adoption cycles (12–18 months on average). Approached as KYC infrastructure, the direct buyers are compliance, claims review, and executive security departments. Adoption cycles, budget line items, and integration paths all differ.
This piece breaks down market size, the pricing and buyer structure of the four major SaaS players, the Hong Kong deepfake incident’s shock, and the Korean KYC 2.0 agenda mapping. Unlike other pieces in the series, this one is roughly 100% market analysis.
Market Size — From $0.6B to $15.1B
Numbers vary by how each research firm defines the market, but the commonality is clear. All classify it as a high-growth category with CAGRs (Compound Annual Growth Rate) above 20%.
| Research Firm | Starting Size | End Size | Period | CAGR |
|---|---|---|---|---|
| Market.us | $0.6B (2025) | $15.1B (2035) | 10 years | 37.2% |
| Mordor Intelligence (Fake Image Detection) | $1.87B (2026) | $7.43B (2031) | 5 years | 31.7% |
| Roots Analysis (AI Detector) | $0.98B (2026) | $7.84B (2035) | 9 years | 26.0% |
| Fortune Business Insights (Deepfake Technology, overall) | $11.18B (2026) | $51.4B (2034) | 8 years | 21% |
| Intel Market Research (AI Deepfake Detector) | $1.2B (2026) | $9.5B (2032) | 6 years | 41% |
Global SaaS averages 13 to 15% over the same period — nearly double. Why the numbers fluctuate by definition: (1) “deepfake detection” narrowly defined vs (2) synthetic content broadly (deepfake + AI-generated images / text) vs (3) including the threat market (deepfake technology itself).
Growth Drivers — Three Data Strands
The growth drivers are reasonably clear. Three data strands point in the same direction.
| Data | Source | Meaning |
|---|---|---|
| Face-swap fraud attempts grew +2,137% over 3 years | Mordor Intelligence (2026) | Absolute threat volume explosion |
| Hong Kong $25M video conference executive impersonation (2024) | Bloomberg, CNN reporting | Single-incident shock effect |
| BFSI KYC workflow upgrade pressure | Mordor Intelligence, Roots Analysis | Direct buyer driver |
Regional split matters. Asia Pacific’s fake-image-detection CAGR runs at 32.5%, the global high. Korea, Japan, and Southeast Asia have mobile finance and instant payments penetration well above global averages, concentrating capital into KYC infrastructure.
The Hong Kong $25M Incident — Market Inflection Point
The January 2024 $25M wire fraud at a multinational in Hong Kong is the deepfake detection market’s biggest inflection point. Incident summary:
| Item | Details |
|---|---|
| Date | January 2024 |
| Victim company | Hong Kong branch of a UK-headquartered multinational |
| Loss | HK$200M (about $25M) |
| Method | Deepfake video impersonating the CFO and five other executives simultaneously in a video conference |
| Employee behavior | ”All meeting participants were familiar faces and voices,” so no suspicion; transfer approved |
| Downstream impact | Global financial institutions immediately strengthened executive security protocols |
The incident’s market impact is clear. Within three months of the incident:
- JP Morgan, HSBC, and Standard Chartered started reviewing mandatory deepfake verification modules in large-transfer approval workflows
- Reality Defender’s revenue doubled quarter-over-quarter (per company announcement)
- 30%+ of Fortune 500 companies started reviewing executive video conference security adoption (Gartner 2024 survey)
- “Deepfake protection” became a new line item in global IT security budgets
A single incident pulled up the entire category’s growth rate. Research firm reports consistently flagging BFSI as “the largest growth segment” has this incident in the background.
Four Major SaaS — Models, Pricing, Buyers Decomposed
Four major SaaS players split the Deepfake Detection category. Each company’s approach, pricing, and buyers differ completely.
Reality Defender — Real-time Video Conference Monitoring
| Reality Defender Profile | Details |
|---|---|
| Founded | 2021 |
| Strength | Real-time video conference stream monitoring, executive impersonation deepfake blocking |
| Pricing model | Enterprise annual subscription (undisclosed; estimated $50K–$500K/year) |
| Main customers | Financial institutions, government, large enterprise C-suite |
| Major round | Series A $33M (2024, led by DCVC) |
| Differentiation | Direct Zoom · Webex · Teams integration, real-time alerts |
Reality Defender specialized in the most visually striking and costly deepfake scenario — impersonating executives on video conferences. The Hong Kong $25M incident became the direct driver. Global financial institution and government adoption accelerated after. Per company announcement, 2025 revenue grew over 4x compared to 2024.
Sensity AI — Forensic-grade Analysis
| Sensity AI Profile | Details |
|---|---|
| Founded | 2018 (started as DeepTrace, rebranded 2021) |
| Strength | Multi-layer deepfake forensics, cloud + on-premise deployment |
| Pricing model | Enterprise + government licensing (analysis-volume based) |
| Main customers | Government agencies, media, financial forensics units |
| HQ | Netherlands (EU market advantage) |
| Differentiation | Multi-modal analysis (image · video · audio · text) |
Sensity specializes in post-hoc forensic analysis — content verification, government investigation, fraud post-incident analysis tracing “where and how this video was synthesized.” EU headquarters becomes an advantageous position once EU AI Act enforcement begins. EU governments and media tend to prefer local solutions, raising adoption likelihood.
Hive — Detection-as-a-Service API
| Hive Profile | Details |
|---|---|
| Founded | 2017 |
| Strength | SaaS API form, content moderation + deepfake detection integrated |
| Pricing model | Per-call + monthly subscription (from $0.0008/image) |
| Main customers | Social networks, gaming, media, SMBs |
| Series E | $50M (2023), $120M+ cumulative |
| Differentiation | Leverages market-leader position in content moderation SaaS |
Hive is the most SMB (Small-Medium Business)-friendly because it’s an API-integration model. It started in content moderation and added deepfake detection as a module. Per-call pricing allows entry proportional to usage volume. Large social networks like Reddit and Quora have automated user-uploaded content verification through Hive.
Truepic — Authenticity Verification (Inverse Approach)
| Truepic Profile | Details |
|---|---|
| Founded | 2015 |
| Strength | ”Controlled capture” — authentication at the moment of shooting |
| Pricing model | Mobile SDK licensing + enterprise |
| Main customers | Insurance claims, court evidence, journalism |
| Series B | $26M (2023, M12 · Microsoft Venture Fund) |
| Differentiation | C2PA founding member, authenticity verification infrastructure |
Truepic runs the inverse approach from the other three. Not “spot the fakes” but “bind the real as real.” The market is smaller than AI detection, but B2B revenue quality is generally rated higher. Insurance claims (Allstate, State Farm, etc.) are core customers, where ROI (Return on Investment) on blocking fraudulent photo/video submissions captures quickly.
One Thing Worth Noting
The four firms get bundled into the same “deepfake detection market,” but their pricing models, buyers, and technical approaches differ entirely. Reality Defender does time-based real-time monitoring (most expensive). Sensity does post-hoc forensics (per analysis). Hive does API integration (most SMB-friendly). Truepic does authenticity verification (the inverse). The 37% average CAGR isn’t single-category growth — it’s the sum of four tracks growing at different rates. A scenario where one company takes the entire category is less likely than four tracks each holding their segment leader position.
The Real Buyers — BFSI Dominates
Mapping where the revenue actually comes from:
| Industry Segment | Deepfake Detection Adoption Driver | Revenue Share (est.) | Representative Adopters |
|---|---|---|---|
| BFSI (Banking, Securities, Insurance) | KYC fraud prevention, executive impersonation blocking, insurance claim verification | 40–50% | JP Morgan, HSBC, Standard Chartered, Allianz |
| Government / Defense | Election deepfake monitoring, fraud investigation, military intelligence verification | 15–20% | US DHS, EU Europol, Korea Police |
| Media / Content | News authenticity verification, content moderation | 10–15% | BBC, Reuters, Meta, TikTok |
| Large Enterprise (Executive Security) | Video conference deepfake blocking, voice fraud prevention | 10–15% | Most of Fortune 500 |
| SMB · SaaS | API-integrated content verification | 5–10% | Reddit, Quora, gaming companies |
BFSI dominates. Research firm reports consistently flag BFSI as “the largest growth segment.” Since face-swap fraud’s +2,137% hits BFSI most directly, budget allocation and solution adoption move fastest there.
The Hong Kong $25M wire fraud became the inflection point. After that, global banks started mandating deepfake verification modules in large-transfer approval workflows. JP Morgan, HSBC, and Standard Chartered have all built in-house deepfake detection infrastructure or adopted external SaaS, per reporting. Insurance is also moving toward mandating synthetic detection modules on claim photos and videos.
KYC 2.0 — Category Integration Flow
What traditional KYC evolves into in the next stage is KYC 2.0.
| KYC 1.0 (existing) | KYC 2.0 |
|---|---|
| ID card photo check | + synthetic photo detection |
| One-time selfie verification | + deepfake voice / face blocking |
| SMS identity verification | + voice synthesis blocking (voice cloning detection) |
| One-time verification at application | + transaction-time continuous verification (continuous KYC) |
| Text / document verification | + synthetic text detection (AI-generated document detection) |
Deepfake detection isn’t entering as a separate category — it’s getting absorbed as a module into existing KYC workflows. Global KYC SaaS (Onfido, Jumio, Veriff) have all added in-house deepfake detection modules or integrated external solutions. This flow’s impact on the Korean market is decisive.
Korea — Direct Connection to the KYC 2.0 Agenda
Korea has world-leading mobile finance and instant payment penetration. Toss, KakaoBank, NICE, and PASS — identity verification and financial rails — function as near-universal infrastructure. Deepfake detection isn’t entering as a new market here; it’s getting absorbed as a KYC 2.0 upgrade module on top of existing infrastructure.
Korean KYC Channel Upgrade Possibilities
| Korea KYC Channel | Current Infrastructure | KYC 2.0 Upgrade Possibility | Impact Timeline |
|---|---|---|---|
| PASS · NICE | SMS + carrier identity verification | Could add deepfake voice blocking module | 2026–27 |
| Toss · KakaoBank | OCR + selfie verification | Could integrate face-swap detection module | 2026+ |
| Insurance claims | Photo / video submission | Could mandate synthetic photo detection | 2026–27 |
| Video conferencing (enterprise) | Zoom · Webex standard | Could deploy executive impersonation deepfake blocking | In progress |
| Government mobile ID | Government 24 + mobile ID | Deepfake blocking + ZKP integration | 2027+ |
The specificity of Korea is that adoption flows through “module addition to existing infrastructure” rather than “new category entry.” User experience barely changes — deepfake detection just gets added one step deeper in the backend. Rather than global SaaS (Reality Defender, Hive) entering directly, Korean identity infrastructure operators like NICE, Dream Security, and Raon Secure licensing global detection engines and integrating them is the more likely path.
Korean Operators Building Detection Too
The flow isn’t entirely global-SaaS dependent. Korean domestic deepfake detection solutions are also growing. Spin-off companies from university research labs (KAIST, Seoul National University) and Korean security startups like SignalLight are trying to differentiate based on local market + Korean-language and Korean-face data. Whether the Korean market settles toward global standards or local standards will be decided over the next 1–2 years.
One Thing Worth Noting
The specificity of the Korean KYC market is that PASS · NICE infrastructure is already laid in. Even if global KYC SaaS (Onfido, Jumio) try to enter Korea, they cannot bypass PASS · NICE — which is tied to identity verification and mobile phones. So even when deepfake detection modules come in globally, they end up layered on top of PASS · NICE as the default structure. This structure could be a good or bad signal for the deepfake detection camp. Good: clear entry barriers mean partnership models stabilize inside the Korean market. Bad: divergence between global and Korean standards adds integration cost per operator.
Closing — Classified as Security, Structured as Infrastructure
Deepfake Detection is classified as cybersecurity, but its revenue structure looks closer to BFSI KYC infrastructure. The buyer is not the security team but the KYC department, executive security, or claims review unit. Adoption isn’t new-category entry but module addition to existing KYC workflows. The Hong Kong $25M incident created a single inflection point for the BFSI camp, and the market got absorbed into the KYC 2.0 flow after that.
Bundled together — all four companies rest on the same assumption. Deepfake detection is not a standalone market but a module being absorbed into KYC 2.0. That’s why direct-sales models matter less for global SaaS than partnership and embedded integration with identity infrastructure operators. In the Korean market, the default path looks like Korean operators (NICE, Dream Security) + global detection engine licensing, rather than direct global SaaS entry.
The next piece closes the series. AI governance heading down a different path from typical regulation patterns — five branches across legislation, academia, market autonomy, identity coupling, and adjacent markets placed on a single plane.
References
- Market.us — “Deepfake Detection Market, CAGR 47.6%”
- Mordor Intelligence — Fake Image Detection Market Report (2026)
- Roots Analysis — AI Detector Market Analysis 2040
- Fortune Business Insights — Deepfake Technology Market Size
- Intel Market Research — AI Deepfake Detector Market Outlook 2026-2032
- CB Insights — Reality Defender, Truepic company profiles
- Reality Defender — official site, Series A announcement, 2025 revenue announcement
- Sensity AI — official site
- Hive — Detection API documentation
- Truepic — Series B announcement
- Help Net Security — “Cyber valuations climb” (2026-02-25)
- Bloomberg, CNN — Hong Kong $25M deepfake incident reporting (2024-02)
- Gartner — 2024 Deepfake Protection Adoption Survey
- Onfido, Jumio, Veriff — internal deepfake detection integration announcements
- Korean NICE, Dream Security, Raon Secure — KYC solution materials
Related Posts
The AI Trust Stack — Content, Personhood, and Agents in Three Layers
As AI-generated content becomes the default, trust infrastructure is splitting into content, personhood, and agent-identity tracks. Five markets, a 5-layer frame, five Big Tech camps, VC flows, and regulation — placed on a single plane.
The Identity Track — From Proof of Personhood to AI Agent Delegation
Proof of personhood and AI agent identity are two layers of the same track. World ID, Passkey, DID adoption curves + Defakto, t54, Indicio funding + payment network entries — the entire identity track in one piece.
SynthID vs C2PA — The Standards War in Adoption Data
The two dominant AI content watermark standards — SynthID and C2PA — adoption broken down by modality, timeline, and camp. Same trust problem at different layers, heading toward coexistence rather than displacement. Plus EU Article 50's operational gaps and compliance gaming scenarios.