Skip to main content
Agentic ScamBank/Account Compromised? Free Official Help

Money or Credentials Compromised? Use Free Official Channels First

1. Call Your Bank Fraud Desk & Police First

If funds, banking passcodes, or wire approvals were compromised, dial the official fraud number printed on the back of your bank card immediately to request a payment recall or account freeze, then report to local police.

2. Report Free to Official AU / US Agencies

In Australia, report free to ACCC Scamwatch (scamwatch.gov.au), IDCARE (idcare.org), and ReportCyber (cyber.gov.au). In the United States, report free to FBI IC3 (ic3.gov) and FTC ReportFraud (reportfraud.ftc.gov).

3. Never Pay "Fund Recovery" Services

Commercial privacy or security subscriptions are strictly preventative; paid tools never recover lost bank transfers or crypto, and agenticscam.com never recommends or links "recovery" agents.

Forensic Acoustic Studio for Synthetic Voice Clones, Video Deepfakes & Rogue Agents

Three seconds of public speech is enough to clone a relative or CFO voice, yet zero-shot neural vocoders and live voice-conversion streams leave measurable 8.0 kHz Nyquist shelves, dead-zero breath dropouts, and 740 ms buffer lag. Inspect live acoustic signatures on the studio bench below and build offline safe-word callback protocols before approving urgent requests.

Live Voice Harmonic Spectrogram & Five-Point Vishing Triage

Toggle between natural human formants and cloned neural vocoder streams across 0 - 16 kHz, then score an active inbound call locally in your browser.

BAND: 0.0 kHz8.0 kHz VOCODER CEILING16.0 kHz
High-Band Phase Cutoff
8.0 kHz Nyquist Vocoder Shelf
Inter-Clause Breath Gaps
0 dBFS dead-zero digital dropouts
Turn-Taking Latency Jitter
410 ms +/- 95 ms (Prompt synthesis)
Micro-Prosody Variance
0.14% over-smoothed prosody

Forensic diagnosis: Sharp horizontal energy shelf above 8.0 kHz caused by 16 kHz neural vocoder upsampling, paired with mathematically silent (-inf dB) inter-clause gaps lacking human inhalation acoustics.

Five-Question Live Vishing Triage

Check every anomaly observed on the call (zero telemetry leaves your browser).

Critical Risk: Hang Up & Call Back Out-of-Band8 / 16 pts (2/5)

Approve zero wires, PayID transfers, or OTP codes on this line. Disconnect immediately, enforce a 15-minute hold, and call back via a pre-saved contact using your Out-of-Band Safe-Word Protocol Card.

Forensic Acoustic Evidence Dossier: Transducer Passbands, Arup HK$200M Bridge & FCC 24-17

Why narrowband phone lines conceal synthetic vocoder ceilings, how multi-person video bridges manufacture group consensus, and how US federal telecom law classifies cloned voices.

Close-up of a broadcast studio microphone capsule representing synthetic voice clone forensics

Channel Passband vs. Vocoder Artifact Visibility

PSTN / Cellular Narrowband (G.711)0.3 - 3.4 kHz

Masks 8 kHz vocoder shelves; rely on turn-taking lag & safe-word callback

HD Voice / VoLTE (AMR-WB)0.05 - 7.0 kHz

Exposes dead-zero digital breath dropouts & metallic RVC phase ringing

WebRTC / Conferencing (Opus Wideband)0.02 - 16.0 kHz

Reveals full 8.0 kHz Nyquist vocoder cutoff shelf & video collar seams

HK$200 Million (US$25.6M) Arup Multi-Person Deepfake Video Call

CONFIRMED INCIDENT · HONG KONG POLICE (FEB 2024) & ARUP DISCLOSURE (MAY 2024)

In early 2024, a finance employee at UK engineering firm Arup's Hong Kong branch received a phishing email citing a confidential transaction. Skepticism vanished after the staffer joined an inbound multi-participant video conference populated by synthetic video and voice clones of the London CFO and multiple recognisable colleagues, resulting in 15 wire transfers totalling HK$200 million across five Hong Kong bank accounts before headquarters was contacted independently.

FCC Declaratory Ruling 24-17: AI Voices Illegal in Unconsented Robocalls

47 U.S.C. § 227 (TCPA) · CG DOCKET NO. 02-278 · ADOPTED 8 FEBRUARY 2024

The U.S. Federal Communications Commission unanimously ruled that AI-generated and cloned voices in outbound telephone calls constitute “artificial” voices under the Telephone Consumer Protection Act, empowering State Attorneys General and the FCC to prosecute callers and block originating voice gateways even when no funds are transferred.

Official Ruling: fcc.gov/document/fcc-makes-ai-generated-voices-robocalls-illegal

Forensic Studio Modules & Consumer Licence Verification

Inspect deepfake call forensics, generate offline verbal safe-word cards, and audit MCP/browser agents, or visit our sister consumer guide Agent Scams (agentscams.com) for plain-English 60-second red-flag checks and official government real estate, travel, and migration agent licence registries.

Studio ChannelTechnical & Regulatory ScopePrimary Audience
Deepfake Call & Video AnatomyArup HK$200M multi-person video call forensics, 8 kHz vocoder shelves, FCC TCPA rulingSecurity awareness leads & finance approvers
Out-of-Band Safe-Word Protocol BuilderPrintable .txt / .md verbal challenge cards, silent duress phrases, 15-minute hold rulesHouseholds, executive assistants & AP desks
Malicious Browser & MCP AgentsIndirect prompt injection, poisoned MCP tool schemas, spoofed OAuth token grantsIT administrators, developers & workspace owners

Optional Preventative Hardening Tools (Never Emergency Recovery)

Paid products cannot reverse an authorized push payment or recover stolen funds; always use the free bank, police, Scamwatch/IDCARE (AU), and IC3/FTC (US) channels at the top of this page first. For proactive pre-incident hardening against reconnaissance and real-time phishing proxies, review these optional tools:

Reviewed by Julian Thorne, Senior Synthetic Media & Autonomous Threat Researcher

Analyses real-time neural voice-cloning artifacts, multi-party video deepfake incidents, and MCP/browser-agent social engineering vectors.

Last reviewed:

Reviewed every 6 months

AI-assistance & editorial check: Research outlining and formatting are assisted by AI tools in an offline draft workflow. Nothing on this site publishes automatically without verification. A named human editor opens and verifies every primary source link, command snippet, and factual statement prior to publication. See our Editorial Policy and Corrections Policy.

Verified Primary Sources

Receive Synthetic-Voice & Autonomous Threat Lab Briefings

Forensic Voice-Clone & Agentic Threat Lab

Store your email for verified forensic advisories when new voice-conversion artifacts, regulatory rulings, or MCP/browser-agent attack vectors are documented. No emails are sent yet.

Do not enter bank, card, passport or password details.
Protected by Cloudflare Turnstile · Isolated per-site D1 (agenticscam-leads).