Estis sees what you can't.
From the Greek ἐστίν — 'it is.' Estis analyses images, video, voice and text for signs of AI — and gives you a sealed, bilingual analysis report you can act on.
Is this you?
That's the trap: anyone can generate images, voices, and video now — and they look real. Estis gives institutions the verdict, the evidence, and the sealed analysis trail the eye alone can't.
The engine
Estis runs on a market-leading detection engine — the same technology relied on by some of the world's biggest newsrooms and studios — wrapped in our bilingual analysis-reporting and credit layer.
The detection engine behind Estis is trusted by The New York Times, Forbes, The Wall Street Journal, Fox, Warner Bros. and Hitachi.
One platform
Detection, reporting, and deployment — one workflow.
Image forensics
AI-generated photos, face swaps, and manipulated images — with quality and tamper signals.
Frame-level analysis
Synthetic video and AI-altered footage flagged before it spreads or enters a case file.
Clone detection
Cloned voices and AI-generated audio identified in recordings, voice notes, and calls — including AI-generated music.
AI-written text
Machine-written essays, statements, and submissions flagged with a confidence score — built for review workflows, not automatic verdicts.
Sealed EN/AR analysis reports
Hash-sealed, timestamped, institution-branded analysis reports — built to be scrutinised by committees and courts.
Independently verifiable
A QR code and reference number let anyone confirm the report itself is a genuine, untampered Estis document — we seal the report, we don't certify the content.
Your systems, wired in
A documented REST API submits media and receives verdicts programmatically.
Hosted where you require
Regional cloud, in-country, or on-premises for data-residency mandates.
From the field
Field teams submit suspect files straight from WhatsApp; original-file guidance preserves analysis quality. Enterprise add-on.
Subject files
Real detection classes from the platform — each subject returns an assessment, a confidence score, and a sealed analysis report.
Who it serves
Verify circulating media and publish findings the public can independently confirm. In-country and on-premises deployment available.
Review signals — never automatic verdicts — for submitted work and admission documents. Designed with educators to protect students and institutions alike.
Verify before publishing. Screen evidence before payouts. Check disclosures before proceedings. Seconds per check, in English and Arabic.
The assessment
A score on a screen is not evidence. Estis reports are built to be scrutinised — by committees, courts, and the public.
Questions
Detection engines return probability scores, not absolute truth — and we say so on every report. High-confidence verdicts are backed by multiple signals, and for high-stakes cases we recommend our human second-review tier. We never market 100% accuracy; anyone who does is selling you risk.
No. Every check is cross-examined by independent detection engines. Images are cross-checked on every submission; audio and video are cross-checked whenever the signals are borderline. When engines disagree, the result is either adjudicated by a further independent engine or reported honestly as uncertain — never forced into a confident verdict.
When the editing tool discloses itself, yes: Google, Adobe, Samsung and Apple tools write AI-edit declarations into the file (IPTC digital source type, C2PA Content Credentials, vendor fields), and Estis reads them and flags the edit prominently on the report. Undeclared local retouches on an otherwise real photo remain a research-grade problem across the whole industry; for those, case files include ELA and noise-analysis indicators for expert review. Originals matter — compressed or messaging-app copies often strip these declarations.
No — Estis is an analysis service, not a certification authority. Every report states probability, methodology, and input quality; the decision remains with you. What we do seal is the report itself: the hash and QR code prove it is a genuine, unmodified Estis document.
For a standalone check, the original file is deleted after analysis. What is kept is the verdict record, the file's cryptographic hash, a small preview image and any evidence images generated from the analysis — enough to verify and read the report later without keeping your media. Workspaces can turn previews off. There are two deliberate exceptions, both under your control: media you attach to a CASE is retained with that case as part of its evidence file, and organizations may enable encrypted retention of every original (Evidence Vault) under their agreement. Retained material is encrypted at rest in object storage in the EU region.
Yes. Files travel over TLS and are processed on encrypted infrastructure, and for a standalone check the original is deleted after analysis — a small preview image and any generated evidence images are kept with the report, encrypted, unless the workspace turns previews off. Two things are retained on purpose: media attached to a CASE stays with that case as evidence, and organizations that need every original kept can enable the optional Evidence Vault. Both are AES-256 encrypted with EU data residency.
Yes. Regional cloud, in-country hosting, and on-premises deployment are available for organizations with data-residency or sovereignty requirements.
Arabic-first, not translated-later: full right-to-left interface, Arabic support team, and every analysis report issued bilingually in Arabic and English side by side.
Annual institutional licenses with included check volumes, or per-check API pricing for claims and screening workflows. Talk to us — we scope a pilot to your case first.
Stop guessing
One check returns a verdict, a confidence score, and a sealed analysis report — in seconds, in your language.
Contact
Tell us about your organization and what you need to verify. We'll arrange a demonstration and a pilot scoped to your case.
Enterprise and institutional engagements only. We respond within one business day.