Research On Forensic Investigation Of Ai-Generated Deepfake Videos In Criminal Cases
1. Rashmika Mandanna Deepfake Video Case (India, 2023-2024)
Facts:
A deepfake video circulated online showing the face of popular actress Rashmika Mandanna superimposed onto another woman’s body in compromising content. The video quickly went viral on social media.
Investigation:
Police seized multiple devices including phones and laptops from suspects linked to the online accounts that shared the video.
Digital forensics experts examined file metadata, device storage, and social media logs to trace the origin of the video.
Analysis revealed that the creator used AI face-swapping software to generate the video from publicly available images of the actress.
Outcome:
The perpetrator, a tech-savvy individual, was identified and arrested.
FIRs were registered under Indian Penal Code sections relating to forgery, identity misuse, and sections of the IT Act for online harassment.
Significance:
Demonstrates how AI-generated deepfakes can be used for reputational harm.
Highlights the role of metadata analysis and device tracing in forensic investigations.
2. Hong Kong Corporate Deepfake Video Call Fraud (2024)
Facts:
An employee of a multinational firm was deceived by a video conference call where the CFO and other executives were impersonated via AI-generated deepfake video and audio. The employee was tricked into transferring millions of Hong Kong dollars to fraudsters.
Investigation:
Forensic teams analyzed the video call logs, capturing inconsistencies in lip-sync, lighting, and facial micro-expressions.
IP addresses, VPN usage, and account credentials were traced to identify the attackers’ locations.
Audio forensic analysis confirmed that voices were synthetic and generated by AI.
Outcome:
The funds were partially recovered through international cooperation, but the perpetrators remain at large.
The case is cited as one of the first large-scale corporate frauds using live AI-generated deepfakes.
Significance:
Highlights the challenge of live or near-live deepfake scenarios.
Shows the need for forensic investigation to include both video and network-level analysis.
3. Ghaziabad Deepfake Extortion Case (India, 2023)
Facts:
A victim was threatened with a deepfake video call showing a police officer in a compromising situation. The perpetrators demanded money in exchange for not releasing the video.
Investigation:
Police traced mobile numbers used for distributing the video and initiating contact with the victim.
Forensic experts examined the original video file and metadata, identifying manipulation artifacts such as unnatural facial movements and inconsistent shadows.
Deleted accounts and social media traces were recovered to link the suspects to the crime.
Outcome:
Perpetrators were arrested for extortion and harassment under the Indian Penal Code and IT Act.
The investigation relied heavily on quick preservation of the digital evidence.
Significance:
Illustrates the use of deepfakes for coercion and extortion.
Shows the importance of rapid forensic response to prevent further distribution.
4. Maryland Audio-Visual Deepfake Harassment Case (USA, 2022)
Facts:
A former high school athletics director created a deepfake video and audio of the principal making offensive statements. The manipulated media was circulated, causing public outrage and reputational harm.
Investigation:
Audio forensics analyzed waveforms, cadence, and spectral fingerprints to detect manipulation.
Video analysis revealed subtle artifacts indicating facial manipulation and inconsistencies in lighting.
Digital evidence from the suspect’s devices confirmed the creation and distribution of the deepfake.
Outcome:
The individual pled guilty and was sentenced for harassment and defamation.
Law enforcement classified this as the first known case of deepfake-based harassment in a school setting.
Significance:
Shows how forensic techniques can detect both video and audio deepfakes.
Demonstrates the legal recognition of deepfakes as criminal tools beyond fraud, including harassment and defamation.
5. UK Online Identity Theft Case Using Deepfake Videos (2019-2020)
Facts:
An individual impersonated someone’s identity online using deepfake videos to convince the victim’s friends and family to transfer funds. The videos featured the victim’s face on another body to simulate personal appeals for money.
Investigation:
Forensic investigators compared the deepfake videos with authentic videos, noting mismatched eye blinking, inconsistent facial expressions, and unnatural movements.
Device metadata, IP logs, and email tracing helped identify the perpetrator.
Outcome:
The suspect was convicted under identity theft and fraud laws.
The case set precedent in the UK for recognizing deepfakes as admissible evidence in financial crimes.
Significance:
Highlights the use of deepfakes in online financial fraud.
Emphasizes forensic video analysis techniques and metadata investigation in establishing authenticity.
Key Takeaways Across Cases:
Deepfakes can be used for fraud, extortion, harassment, and reputational harm.
Forensic investigation relies on metadata analysis, device tracing, video/audio artifact detection, and social media forensic analysis.
Live or near-live deepfakes (video calls) present unique challenges requiring network and call-log investigation.
Courts are beginning to accept deepfake evidence if forensic methodology is robust and documented.
RELATED Blog
comments