Check if it's a deepfake.
Detect deepfakes in audio, images, and video — free, from Resemble AI. Drop a file and get a verdict, plus the reasoning behind it.
Select an audio, image, or video file.
Opt in to depth analysis on the why behind the results.
Drop your audio file here
MP3 · WAV · M4A · FLAC · max 50.0 MB
Deepfake incidents Resemble is built to stop.
The strongest signal on this site is the public record: real voice-clone scams, synthetic images, video-call impersonation, election disinformation, and platform abuse. The database refreshes as new verified incidents are added, so the homepage stays grounded in what attackers are doing now.
Arup (global engineering firm)
A finance employee at engineering firm Arup wired $25.6M after a multi-person video call where every participant except them was a deepfake.
New Hampshire voters / Biden campaign
Cloned audio of President Biden told New Hampshire Democrats not to vote in the primary.
Finance worker in Hong Kong (company's CFO and other executives)
A finance worker in Hong Kong transferred $25 million to scammers after a deepfake video call impersonated the company's CFO and other executives.
Fresh examples for deepfake fraud, AI voice scams, and synthetic media abuse.
Each incident links to cited sources and a breakdown of where detection, provenance, watermarking, or workflow controls could have helped. This is why the detector exists: not as a demo, but as a public front door into Resemble's deepfake intelligence stack.
S Jaishankar and Sergey Lavrov
Viral deepfake videos falsely depicted Indian and Russian foreign ministers making controversial statements about international relations and diplomatic.
Bubblemaps
Bubblemaps refutes claims of a secret consulting deal after an X user circulated a fabricated contract allegedly generated by AI to damage the company.
Donald Trump
The Iranian Embassy in India used an AI-generated video on X to mock Donald Trump following Republican losses in the US midterm elections.
Pet owners Stephanie Young, Cam Chauhan, and Sarah Robinson
Scammers are using AI-generated images of missing pets to deceive desperate owners into paying fraudulent rewards and issuing threats of harm
Hany Farid
Computer scientist Hany Farid was targeted by an attacker who used a voice clone to impersonate him during a phone call to an attorney regarding a sensitive.
Sergey Lavrov and S Jaishankar
Authorities debunked AI-generated deepfake videos targeting Indian and Russian foreign ministers in an attempt to disrupt diplomatic relations before the SCO.
From suspicious links to live deepfake meetings.
Most incidents start in the places teams already work: browser tabs, shared media, video calls, and meeting rooms. These two Resemble products are the clearest next steps from the public detector.
Deepfake Detection Chrome Extension
Check suspicious audio, video, and images while browsing the web. Useful for journalists, analysts, trust-and-safety reviewers, and anyone validating media before sharing it.
- Verify media in the browser
- Route suspicious content into deeper analysis
- Useful for incident triage and open-source research
Deepfake Detection for Meetings
Bring deepfake detection into Zoom, Microsoft Teams, Google Meet, and Webex workflows where executive impersonation and live-call fraud can happen.
- Built for live meeting risk
- Supports Zoom, Teams, Google Meet, and Webex
- Helps teams respond to video-call impersonation
More than a detector.
A full trust layer for AI media.
HIBP for breaches. VirusTotal for malware. detectdeepfakes.com is the trust utility for synthetic media — free tools for the world, powered by the same stack Resemble AI deploys to banks, telcos, and newsrooms.
Paste any URL — we extract the text and images, run every asset through our detectors, and return one unified risk report.
Check any audio, image, or video for Resemble's multimodal watermark. C2PA and SynthID layering in development.
Upload a voicemail. We flag AI-generated voice and map it against 20 known scam scripts — the engine banks and telcos use in production.
Every major deepfake incident on record, documented with cited sources. Free to cite for journalists and researchers.
Drop our detector into your own site or newsroom. One script tag. Free for non-commercial use.
Detection is one layer. Incidents show where it fits.
For fraud teams
Use the audio and video detectors to triage suspicious calls, submitted recordings, and executive impersonation evidence before money moves.
For newsrooms
Verify viral media against detector signals, provenance clues, and the incident database before treating a clip as evidence.
For platforms
Map incident patterns into trust-and-safety workflows: AI images, synthetic profiles, voice cloning, and coordinated abuse campaigns.
For developers
Move from one-off checks to API-based detection when uploads, calls, or user-generated media need continuous screening.
A free detector backed by real attack intelligence.
What the detector checks
Deepfake detection works best when several signals are considered together. Audio files can expose vocoder fingerprints, unusually regular timing, or cloned-speaker artifacts. Images can reveal diffusion-model texture, inconsistent camera noise, manipulated faces, missing provenance, or repeated compression patterns. Video adds temporal evidence: identity drift, lip-sync mismatch, unstable lighting, and frame artifacts that are hard to see by eye.
Why the incident database matters
Search results for deepfake detection are full of generic tools. The difference here is the incident record underneath the interface. Each verified case shows the attack pattern, the affected victim or organization, the media type, and the public sources behind the report. That makes the homepage useful for people searching for a free deepfake detector, deepfake fraud examples, AI voice scam evidence, or a cited database of synthetic-media abuse.
When to move from a free check to Resemble AI
A one-off upload is enough for a suspicious voicemail or viral image. Organizations need a different layer: API-based screening, media-intelligence workflows, watermarking and provenance checks, model updates, and reporting that fits fraud, trust-and-safety, newsroom, or compliance operations. That is the role of Resemble AI behind this public detector.
Frequently asked questions
+How does the free deepfake detector work?
Upload audio, image, or video and the detector looks for synthetic-media signals such as voice-clone artifacts, diffusion fingerprints, face-swap traces, and inconsistent provenance signals.
+Why feature deepfake incidents on the homepage?
The incident database shows the real attacks behind the tool: voice-clone fraud, election robocalls, synthetic images, video-call impersonation, and other AI-media abuse cases.
+Can Resemble prevent every incident listed here?
No single product prevents every attack. Resemble provides detection, watermarking, verification, and intelligence layers that help organizations identify and reduce deepfake risk.
+Is the incident database updated?
Yes. The Deepfake Incident Database is maintained as a continuously updated public reference with cited sources and stable incident URLs.
Need deepfake detection inside your product or workflow?
Resemble AI gives teams the detection API, watermarking and provenance checks, media intelligence, and operational workflows needed to respond to the same attack patterns tracked in the incident database.