MAS SIMPLE SA USER. MAS MAHIGPIT SA ILALIM.

Scam Ba ’To? was built for the few seconds before a risky tap. Hindi mo kailangang maging cybersecurity expert. Kailangan mo lang ng malinaw na sagot bago ka mag-click, magbayad, o magbigay ng impormasyon.

Hindi ito “AI na nanghuhula.” May context analysis, evidence checks, verified registry data, action-risk evaluation, at safety gates bago ilabas ang verdict.

Hindi lang namin tinatanong kung mukhang scam ang message.

Sinusuri namin kung ano ang pinapagawa nito sa iyo, sino ang sinasabi nitong nagpadala, saan ka nito dinadala, at kung may ebidensyang sumusuporta sa claim.

Paano ito gumagana

Mula sa message hanggang sa verdict.

Ang interface ay simple. Sa ilalim nito, maraming independent checks ang nagkikita bago maglabas ng sagot.

Action Risk

Hindi porke may “OTP” ay scam na.

Context matters. Maraming signal ang tinitingnan nang sabay bago magdesisyon.

01Pagkakakilanlan
02Hinihinging aksyon
03Link at domain
04Pressure / pananakot
05Pera at account
06Official evidence

Decision model

Interpretation helps.
Evidence decides.

Ang system ay gumagamit ng multilingual semantic parsing at factor-based intelligence para maintindihan ang context. Hindi isang language model ang nag-iisang “judge.” Ang verdict ay dumadaan sa technical evidence, verified registry data, action-risk evaluation, at safety rules.

01Semantic interpretationunderstand context & requested action
02Evidence & registrycheck domains, entities & official guidance
03Action Risk Engineevaluate consequence and behavioral pressure
04Safety policyconstrain unsafe classifications
05Human-readable verdictred · yellow · green

Green Eligibility

Mas mahirap maging Green.

Ang kawalan ng obvious keyword ay hindi sapat. Bago maging Green, kailangan hindi ma-trigger ang mga safety condition na maaaring magtago ng high-consequence risk.

Candidate Green
Identity
Action risk
Evidence
Safety policy
Green eligible
Green does not mean “guaranteed safe.” Ibig sabihin lang nito, walang sapat na evidence sa message para magbigay ng mas malakas na warning.

Testing

Hindi lang sa madaling examples tinest.

Ang current release candidate ay dumaan sa frozen adversarial, historical, screenshot, policy, at production-hardening gates.

500/500

Frozen adversarial benchmark

100/100Attack #5 safety-policy corpus
191/191Core / historical batch
499 pass1 false GreenRELEASE BLOCKED

Isang reproducible high-consequence scam na ma-classify bilang Green ay sapat para i-block ang V1.0 at gawing permanent regression test.

Gold Registry

Real organization ≠ tunay ang sender.

Claim sa message“GCash kami.”

Claim lang ito. Hindi pa authentication.

Verified registrygcash.com

Official domain evidence na hiwalay sa suspicious message.

Software design

Production architecture.

Presentation

Filipino-first, safety-oriented interface

FastAPI API

validation, privacy sanitization, orchestration

Intelligence

multilingual parsing, identity and URL analysis

Action Risk

factor-based consequence evaluation

Evidence

Gold Registry and official guidance

Decision & Safety

risk fusion, critic, Green Eligibility

Current build: v1.3.6-rc2-actionrisk · Architecture update: August 2026

Limitasyon

Hindi ito magic shield.

Hindi pinapatunayan ng Green result na tunay ang sender. Hindi rin nito mapapalitan ang bangko, government agency, telco, o law-enforcement authority.

Para sa pera, account access, OTP, password, o personal information, dumiretso sa independently verified official app, website, o contact channel.

Professor John Clement Escobañez, MSIT

Professor · College of Computing and Information Technologies · National University – Manila

John Clement Escobañez is an information technology professor whose work sits at the intersection of applied artificial intelligence, decision intelligence, systems integration, digital governance, and human-centered computing. His teaching and project work span web systems, enterprise and systems architecture, multimedia technologies, and the practical use of emerging technologies in real-world settings.

He develops technology with a strong emphasis on usability, explainability, and decision support—especially for people who should not need technical expertise just to make a safer choice. Scam Ba ’To? follows that approach: complex checks underneath, a direct answer on the surface.

FIELDInformation Technology
FOCUSApplied AI · Decision Intelligence · E-Governance
TEACHINGWeb Systems · Systems Integration · Multimedia Technologies

Institutional affiliation is shown for professional identification and does not by itself imply official institutional endorsement of this independent project.

Sa dulo, simple lang.

Bago mag-click,
mag-check muna.

Kung mapigilan nito kahit isang maling click, maling transfer, o maling pagbibigay ng impormasyon, ginagawa nito ang trabaho nito.

Suriin ang isang message →