Shell WalkerIntermediate10 daysFree
AI Deepfake Detection and Digital Forensics Masterclass
In ten days you will go from a suspicious file to an evidence-backed verdict you can defend in a report.
Taught by Mr Elite 0 of 10 days live A new day every day
This course teaches you to examine images, video and audio and decide, with evidence, whether they are genuine, edited or machine-generated. I will show you the tools working practitioners use: ExifTool, ffmpeg, Python, Audacity and the C2PA command line tool. We work on files you create or that are freely licensed, on your own machine, plus the SecurityElites practice labs where one fits.
Each day is one lesson of 45 to 75 minutes, including a mission that produces something checkable, such as a screenshot, a note or a short write-up. We start with setup and evidence handling, move through images, video and voice, then finish with provenance, incident investigation and a capstone case. I will be honest about the limits too. No single detector gives you a verdict, so you will learn to combine weak signals into a conclusion you can explain.
By the end you can
- Handle suspect media as evidence, with hashes, working copies and a simple chain of custody.
- Read and interpret metadata and file structure to find signs of editing or generation.
- Examine images for generation and manipulation artefacts and judge how strong each signal is.
- Analyse video frame by frame and check audio and video for consistency.
- Compare synthetic and genuine speech with spectrograms and basic features.
- Verify content credentials and provenance with C2PA tooling.
- Run an incident investigation and write a clear forensic report with stated confidence.
Who this is for
- Security analysts, SOC staff and IT responders who need to assess suspicious media.
- Ethical hackers and students who are comfortable in a terminal and want a forensics skill set.
- Anyone handling fraud, impersonation or disinformation reports who must give a defensible answer.
Before you start
- Comfort using a Linux, macOS or WSL terminal, including running commands and editing files.
- A basic grasp of how files, the web and networks work.
- Some Python reading ability helps, but I explain every script we run.
- Linux, macOS or Windows with WSL
- ExifTool
- ffmpeg and ffprobe
- MediaInfo
- Python 3 with Pillow, NumPy, librosa and matplotlib
- Audacity
- espeak-ng
- c2patool
Syllabus
One day at a time. Finished days get a tick. Days that are not out yet show when they unlock.
Module 1 Evidence, Metadata and Image Forensics
Handle media as evidence and judge whether an image is genuine, edited or generated.
- 1Orientation, Setup and Your First Evidence HashBy the end of today you can install the toolkit and record a verified hash of a suspect file.Coming soon
- 2Metadata and File Structure ExaminationBy the end of today you can extract metadata and file structure details and explain what they do and do not prove.Coming soon
- 3Spotting AI-Generated ImagesBy the end of today you can examine an image for generation artefacts and rate how strong each signal is.Practice | Lab: AI Visual Prompt Injection | Coming soon
- 4Apply It: Image Triage CaseBy the end of today you can triage a set of suspect images and review the Day 1 to 3 methods in one workflow.Review | Lab: Real or Generated? | Coming soon
Module 2 Video and Voice Analysis
Analyse video and audio for signs of synthesis and know where detection breaks down.
- 5Deepfake Video AnalysisBy the end of today you can break a video into frames and streams and look for inconsistencies that suggest manipulation.Coming soon
- 6Voice Cloning DetectionBy the end of today you can compare genuine and synthetic speech using spectrograms and simple audio features.Practice | Coming soon
- 7Apply It: Mixed Media Case and Detector LimitsBy the end of today you can analyse a video with audio, review earlier methods and explain how detectors can be evaded.Review | Lab: ML Evasion | Coming soon
Module 3 Provenance, Investigation and Capstone
Verify where media came from, run a full investigation and report it.
- 8Media Provenance and Content CredentialsBy the end of today you can read and verify C2PA content credentials and say what a missing credential means.Coming soon
- 9Incident Investigation and ReportingBy the end of today you can run a media investigation from report to written findings with stated confidence.Practice | Coming soon
- 10Capstone Case File and Where to Go NextBy the end of today you can investigate an image, a video and an audio clip as one case and deliver a defensible written report.Capstone | Coming soon
The final project
You receive a case file containing an image, a short video and an audio clip tied to a fictional impersonation incident, and you investigate all three. You finish with a written forensic report that states your findings, your confidence in each and your recommended response.
Questions people ask
Do I need to create deepfakes to take this course?
No. You work with your own photos and voice, a plain synthetic voice from espeak-ng, and freely licensed samples. I never ask you to imitate a real person.
Can a tool tell me for certain that media is fake?
No, and I will show you why. Detectors give probabilities and can be fooled, so you learn to combine metadata, artefacts, provenance and context into a judgement you can explain.
Is everything free to run?
Yes. Every tool listed is free and runs on your own machine, and the labs run in your browser. You do not need a paid account for anything.
How long does each day take?
Plan on 45 to 75 minutes, including the mission. Take your time with the missions, because each day builds on the one before.