Problem Statement
CICC says AI made love scams “mas organisado, mas nakakaloko” — foreigner Investor lures to fake forex, Serviceman targets widows, Seducer extorts with nudes, Slow Burn builds months of trust. Victims are blamed, so they don’t report to 1326 until money is gone. Families see red flags but can’t intervene kindly.
Proposed Solution
Paste chat or forward screenshots: LoveScamLens profiles the playbook (“Investor — 84%: luxury lifestyle + forex link + urgency”), highlights 3 quotes as proof, and gives a gentle next-step script for the victim and family (“Hindi ka tanga — organisado sila. Gawin ito ngayon”). Anonymized reports feed a live Investor-coin dashboard for CICC.
Target Market
- Primary users: Solo parents, OFW families, seniors, students new to dating apps.
- B2G: CICC, PNP-ACG, LGU VAWC desks.
How AI Is Used
- Playbook classifier: LLM trained on CICC’s 6 profiles; keyword rules miss Slow Burn grooming.
- URL + persona check: Domain age + reverse-image hints without doxxing.
PLDT / Smart / ePLDT / Rokid Angle
- Smart: SMS forwarding short code; intel feeds spam filters.
- ePLDT: Hosts trends dashboard in-country for CICC.
Demo Scenario (3 minutes for judges)
- Paste mock Investor chat (“I’m a crypto trader in SG, invest ₱10k…”).
- Verdict INVESTOR 84% with 3 highlighted lines appears in Taglish.
- Show gentle family script + 1326 steps.
- Dashboard shows mock Investor-coin spike in Region IV-A.
- Close: “Hindi kahinaan ang magmahal.”
30-Hour Build Plan
| Hours | Task | Role |
|---|---|---|
| 0–3 | Mock chats from CICC profiles | All |
| 3–12 | Profiler prompts + Taglish tone | AI lead |
| 3–10 | URL check + anonymizer | Backend |
| 10–20 | Chat + dashboard UI | Frontend |
| 20–26 | Sensitivity review | All |
| 26–30 | Pitch, disclosure | PM / Pitch |
Suggested stack: Chat UI + LLM scam profiler, URL checker, SMS forward, Postgres trends. Public data / APIs: CICC Feb 2026 profiles,hold; URLhaus, WHOIS (mock for demo).
Risks & Privacy Notes
- Victim safety: Non-judgmental tone, no victim-blaming; crisis resources listed. Demo chats synthetic; no real chats or nudes into AI tools.
- Privacy: Chats auto-redacted, 90-day retention, consent-first per Data Privacy Act.
- Abuse: Rate-limited so scammers can’t test scripts.
AI Use Disclosure (draft)
Uses LLM profiling plus URL checks. Coding assistants helped with UI. All demo chats synthetic and reviewed by the team for sensitivity.
Rubric Scoring
| Criterion | Weight | Score (1–5) | Why |
|---|---|---|---|
| Innovation & creativity | 25% | 4 | CICC-profiled Taglish profiler is new |
| Problem–solution fit | 25% | 5 | AI love scams surging per CICC |
| AI application & value | 20% | 4 | Grooming detection needs LLM |
| Feasibility | 15% | 5 | Very buildable |
| Impact & scalability | 15% | 4 | Nationwide via Smart |
| Weighted total | 4.40 |