Intimacy Risk Infrastructure

TORVEN

The missing infrastructure layer for romance scam prevention — training data that doesn't exist elsewhere, and detection that works over weeks, not seconds.

White-label infrastructure for Intimacy Risk Intelligence. Enabling platforms to detect and prevent romance scams before financial or emotional harm occurs.

Express Interest View Products
$672M
US Losses 2024
FBI IC3
~456%
Cumulative Growth
FTC 2019–2023
0%
Detection Rate
LlamaGuard 3 on Romance Scams
02 — The Problem

Romance scams are the fastest-growing fraud category. Current detection achieves 0%.

FTC-reported losses grew from $201M to $1.14B between 2019 and 2023 — approximately 456% cumulative growth. Median victim loss is $10,000. AI-powered scams now scale with synthetic personas and LLMs enabling single operators to manage 50+ victims simultaneously. No infrastructure exists to stop them at the conversational level.

01
$672M
FBI IC3 Reported Losses (2024)
17,910 complaints. Global losses estimated 3–5x higher due to systemic underreporting and victim stigma.
02
0%
Detection Rate
LlamaGuard 3 achieved 0% true positive rate on romance scam conversations (Gressel et al., 2025). Messages appear friendly — every scanner misses them.
03
Zero
Public Training Datasets
No commercial-scale dataset of long-term romance scam conversations exists. Privacy barriers, evidence destruction, and victim stigma prevent collection.
04
$14M
Match Group FTC Settlement
August 2025 — deceptive practices and profiting from messages by flagged accounts. FTC proceedings revealed over half of incoming messages from fraudulent accounts in certain periods.
03 — Dual Product Infrastructure

Two products. One compounding advantage.

Data Farm solves the cold-start problem — training data that doesn't exist elsewhere. Sentinel deploys detection where scams happen — inside conversations, over time. Together they form a closed-loop flywheel.

Phase 1 — In Development
Data Farm
Psychology-accurate synthetic romance scam conversations for training fraud detection AI — without exposing real victim data.
  • Full temporal arc: initial contact → trust building → exploitation → extraction
  • Multi-scenario: military, crypto, inheritance, medical emergency variants
  • Configurable victim personas: demographics, vulnerability factors, communication styles
  • Every message labeled: scam stage, manipulation tactic, risk indicators
  • GDPR-safe synthetic data — no real victim conversations required
€1,499 — €9,999 per dataset (1,000–10,000 conversations) · Gross margin 70–92%
For: Cybersecurity vendors · AI/ML training teams · Dating platform moderation · Academic research · Financial institutions
Request Early Access →
Phase 2 — Planned
Sentinel API
Invisible white-label infrastructure that runs underneath your platform — detecting temporal manipulation patterns over weeks and months, not scanning individual messages.
  • Behavioral pattern analysis across full conversation lifecycle
  • Temporal modeling: tracks relationship progression and acceleration patterns
  • Risk scoring + reason codes + configurable intervention thresholds
  • API-first: SDK integration for dating, messaging, social platforms
  • Privacy-first: edge-isolated inference, no content retention
€0.04 — €0.15 per conversation · ACV €40K–200K by platform size · 80%+ gross margin
For: Dating platforms · Messaging apps · Telcos/ISPs · Social media · AI companion platforms
Register Interest →
04 — Closed-Loop Infrastructure

The flywheel effect.

Data Farm and Sentinel are not independent products. They form a self-reinforcing system. Competitors building only detection or only data cannot close this loop.

Data Farm
Generates synthetic conversations using behavioral pattern library
→Trains
Sentinel API
Detects real-world scam patterns in platform conversations
Sentinel discovers new patterns → improves Data Farm generation → trains better models → improves detection → compounding advantage
05 — Who It's For

Built for teams who need romance scam capabilities but can't build the data infrastructure alone.

If you operate a platform with interpersonal communication, provide fraud detection services, or train AI safety models — and you don't have a 10-person ML team to build a romance scam conversation generator — TORVEN exists for you.

Data Farm
Content Moderation & Trust & Safety
You moderate dating platforms but train on real data with GDPR exposure. Our synthetic datasets replace that risk with labeled, configurable conversations.
Data Farm
White-Label Dating Software
Your clients run dating sites and need scam protection as a feature. Integrate TORVEN data to power AI detection inside your platform product.
Data Farm
Security Awareness Training
You cover phishing and BEC but not romance scams. Our conversations become realistic training scenarios for your next module — the fastest-growing social engineering category.
Sentinel API
Dating Platforms Under Regulatory Pressure
FTC enforcement, EU AI Act, Senate hearings — you need infrastructure-grade compliance. Sentinel provides white-label detection your Trust & Safety team can deploy without ML expertise.
Sentinel API
Messaging & AI Companion Platforms
Wherever strangers form emotional connections online, romance scams follow. Sentinel's API integrates into your existing communication infrastructure.
Data Farm + Sentinel
Fraud Detection Vendors
You serve dating platforms but lack romance scam capability. TORVEN becomes your plug-in module — expand coverage without R&D investment. Revenue share partnerships available.
06 — Why Now

Three converging forces create a time-limited market window.

The infrastructure gap is open. The first entrant with production-grade temporal detection captures the market.

01
Regulatory Catalyst
Match Group $14M FTC settlement (Aug 2025). Senate hearings on dating platform safety (Sep 2025). EU AI Act compliance requirements. UK PSR mandatory APP fraud reimbursement (Oct 2024). Platforms need infrastructure, not features.
02
AI Threat Acceleration
Deepfake tool trading grew 223% (2023–2024). LLMs enable single operators to manage 50+ victims simultaneously with personalized conversations. Detection systems need equally sophisticated training data.
03
Infrastructure Vacuum
Consumer solutions exist (Norton per-message scanning). But no infrastructure-grade solution addresses long-horizon conversational detection — the temporal manipulation that unfolds over weeks and months.
04
Technology Readiness
LLM advances enable synthetic data generation at viable cost. Behavioral AI models achieve production-grade accuracy. API-first integration reduces adoption friction. The building blocks exist — the application doesn't.
07 — Expression of Interest

Join the early access program.

Submit a non-binding expression of interest. Upon submission, a formal Letter of Intent (PDF) will be generated and downloaded to your device.

Non-Binding Letter of Intent
Data Farm — Early Access

Express your interest in accessing synthetic romance scam conversation datasets for training, research, or compliance.

Intended for:
  • Content moderation & Trust & Safety teams
  • Cybersecurity vendors training fraud detection models
  • AI/ML teams building safety classifiers
  • White-label dating software providers
  • Security awareness training companies
  • Academic institutions researching online fraud

Non-binding pursuant to standard commercial LOI practices. No commitment to purchase implied. Either party may withdraw at any time. Books & Mirror Inc. will treat all information as confidential.

Generating LOI document...
✓
LOI Submitted & Downloaded

Your non-binding Letter of Intent has been recorded and a PDF copy downloaded. We will contact you with early access details.

Non-Binding Letter of Intent
Sentinel API — Early Access

Express your interest in integrating real-time fraud detection infrastructure into your platform when available.

Intended for:
  • Dating platforms seeking compliance-ready fraud prevention
  • Messaging apps with interpersonal communication
  • Social media platforms with direct messaging
  • Telecom / ISP providers
  • AI companion and voice avatar platforms

Non-binding pursuant to standard commercial LOI practices. No commitment to purchase, license, or integrate implied. Either party may withdraw at any time. Books & Mirror Inc. will treat all information as confidential.

Generating LOI document...
✓
LOI Submitted & Downloaded

Your non-binding Letter of Intent has been recorded and a PDF copy downloaded. We will contact you with pilot program details.

Contact

Let's talk.

Whether you are a platform seeking fraud prevention infrastructure, a fraud detection vendor exploring partnerships, or a researcher studying online fraud — we are ready to talk.

Partnership Inquiry Investor Inquiry
TORVEN · Books & Mirror Inc. · Delaware C-Corporation · Est. 2025