AI-powered forensic platform for comprehensive mobile threat analysis. Designed for security professionals and forensic examiners, LUMA detects state-sponsored spyware, commercial surveillance tools, and Zero-Click / One-Click attack indicators on iOS and Android devices.
One product, five license tiers. Same forensic engine, scaled to your scan volume.
Multi-layer forensic analysis powered by AI and deterministic computation
172,500+ indicators of compromise from Amnesty Tech, MVT, AssoEchap, and proprietary databases covering 87 actors (61 state-sponsored + 26 commercial stalkerware vendors).
AI-powered behavioral analysis detecting surveillance patterns, sensor cascades, and covert data exfiltration.
Evidence-based threat probability scoring with weighted assessment and scenario-driven analysis framework.
Court-admissible multi-language reports with full evidence chain, methodology documentation, and expert conclusions.
Deep analysis of sysdiagnose, bugreports, crash logs, network traffic, permissions, and app behavior on both platforms.
13 threat scenarios including Zero-Click exploits, night surveillance, sensor cascades, forensic extraction, telco-side SS7, profile-based MITM, and per-actor signature detection.
From device connection to forensic report in three stages
Connect the device via USB or receive a diagnostic file remotely through a secure QR-based upload link. iOS sysdiagnose and Android bugreport capture system-level logs without accessing personal data.
LUMA runs a multi-layer scan: IOC matching against 172,500+ indicators, behavioral pattern detection across sensors and network, process anomaly scoring, and 13-scenario threat modeling.
AI synthesizes findings into a court-admissible forensic report with Bayesian evidence weighting, full methodology documentation, and a clear verdict with confidence level.
Scan a phone in another city, country, or hostile environment without ever touching the device. Same engine, same scenarios, same court-admissible report.
No jailbreak, no root, no specialized hardware. iOS sysdiagnose and Android bugreport are generated natively by the device. The client never relinquishes physical custody.
Pattern-based detection of real-world attack techniques, from zero-click exploits to physical extraction
System daemon crash followed by covert sensor activation. Consistent with iMessage/WhatsApp zero-click exploitation.
Sensors active during 01:00-06:00 with network data exfiltration. State-sponsored surveillance pattern.
Sequential activation of mic, camera, GPS, and Bluetooth within minutes. Programmatic intelligence collection.
Regular-interval server connections with asymmetric upload volume. Active command-and-control communication.
PT_DENY_ATTACH, truncated backtraces, SIGABRT in security processes. Hallmark of state-sponsored tools.
Daemon crash followed by anomalous process behavior. Privilege escalation via daemon restart vulnerability.
WiFi or Bluetooth anomaly indicating close-range attack vector. Hotel, airport, and meeting scenarios.
Detection of Cellebrite, GrayKey, MSAB, and Oxygen tools. Identifies unauthorized physical forensic acquisition.
Configuration profile with root CA + VPN/proxy enabling HTTPS interception. iCloud / OAuth token theft pattern.
Indirect symptoms of SS7 / Diameter network-level interception. Identifies victim-side evidence of telco exploits.
Detection of suspicious Global Title roaming activity consistent with documented SS7 attack infrastructure.
Combined cellular, WiFi, and configuration profile anomalies. Multi-vector targeting pattern.
Cross-correlated detector evidence matching specific actor TTPs (Pegasus, Predator, Candiru, Hermit, others).
Local-only AI engine. Scans never leave the device, never reach a third-party cloud, never depend on connectivity.
LUMA is designed to meet the highest standards required by government agencies, law enforcement, and enterprise security teams.
The examiner workstation runs the entire forensic pipeline locally - including the AI engine. The model size and inference speed are bounded by GPU memory. Pick a tier that matches your scan volume.
| Hardware tier | Clean iPhone scan | Compromised iPhone scan |
|---|---|---|
| Mac M-series (any) | 1-2 min | 2-3 min |
| Windows, 24 GB+ VRAM | 1-2 min | 2-4 min |
| Windows, 8-16 GB VRAM | 2-3 min | 4-7 min |
| Windows, 4-8 GB VRAM | 3-5 min | 7-12 min |
| Windows, integrated / under 4 GB VRAM | 5-10 min | 12-25 min |
First-run AI model download is approximately 17 GB (one-time, per workstation). Air-gap mode is supported after initial setup. Scans never leave the device - the entire forensic pipeline runs locally.
Available for Windows and macOS. Licensed for authorized security professionals.
Contact us to schedule a live forensic demo or discuss deployment for your organization.
Contact Us