Cyberette launches deepfake forensics platform to show where media was manipulated

Amsterdam-based media forensics startup Cyberette launched a deepfake detection platform on September 30 for fraud, identity and investigative teams that need more than a verdict on whether an image, video or voice recording is fake.

The platform flags regions of images and timecodes in videos that appear to have been manipulated and compiles its findings into forensic reports. For identity verification, it can connect to existing systems through an API to detect face swaps, synthetic faces, cloned voices and injection attacks.

“A real-or-fake score is where the work starts, not where it ends,” said Julia Jakimenko, Cyberette’s co-founder and CEO. “Whether you’re approving a new customer or taking evidence to a judge, you need to show what was changed, where, and how strong the evidence is.”

Two jobs: stopping fraud and documenting it

Cyberette is targeting two related but different workflows.

In identity verification, speed matters. The platform analyzes incoming media for signs of manipulation and Cyberette says it can return results in under two seconds for live scenarios. South Korean identity and payments company Crosshub has also partnered with Cyberette.

For investigators, the emphasis shifts toward documenting what happened. The system analyzes facial geometry, pose and motion for inconsistencies, produces heatmaps identifying suspicious areas of a frame, and examines audio for patterns associated with synthetic speech. It also analyzes file metadata and C2PA content credentials.

Jakimenko says Cyberette can provide information about manipulation patterns, possible model families used and approximate dates, along with other provenance indicators where available.

Customers can run the platform in Cyberette’s cloud or on their own infrastructure, an important consideration for police, defence and other organizations that may be unable to send sensitive evidence to a third party.

Cyberette currently reports 98 percent detection accuracy across image, video and audio. The figure is a company performance claim and has not been independently benchmarked. Cyberette says its platform is available now.

Deepfake detection checks forecast to surge

Cyberette is entering the market as the scale of deepfake-enabled fraud accelerates.

Biometric Update and Goode Intelligence’s Deepfake Detection Market Report forecasts billions of annual deepfake detection checks as organizations respond to synthetic identity, account takeover and other AI-enabled fraud. The report describes deepfake detection as an increasingly important layer in identity verification, authentication and fraud prevention, while warning that a fragmented vendor landscape makes choosing between technologies difficult.

Demand for detection is growing rapidly alongside the threat. The report forecasts total global voice and face deepfake detection checks at more than 6 billion annually in 2026, rising to 9.9 billion in 2027. Global revenue from voice and face deepfake detection is forecast at approximately $3 billion in 2026 and $4.95 billion in 2027.

Regulators are also putting more emphasis on identifying AI-generated content. Article 50 of the EU AI Act includes transparency obligations around certain AI-generated content and deepfakes.

Jakimenko argues that transparency requirements have an obvious limitation when dealing with criminals.

“Labels help when people follow the rules,” she said. “The people creating fraudulent or abusive deepfakes won’t. That’s where forensics comes in.”

That logic underpins Cyberette’s combination of provenance and detection. Where media carries C2PA credentials, the platform can examine information about its origin and editing history. Where authenticated provenance is absent, forensic analysis provides another way to look for manipulation.

Specialists compete with established identity players

Cyberette enters a fragmented market that includes biometric and identity providers, voice specialists and dedicated deepfake detection companies. The Biometric Update and Goode Intelligence report profiles vendors including Pindrop, Reality Defender, Paravision, ROC and Mobai, while Amsterdam-based Sensity also targets forensic deepfake detection.

Cyberette sits toward the specialist end of that spectrum. Its focus is multimodal media forensics across image, video and audio rather than a broader identity platform. That means Cyberette is competing both with specialist deepfake vendors and with larger biometric and fraud platforms adding detection to existing identity workflows.

For a specialist, integrations become particularly important. Rather than replacing an organization’s identity, fraud or forensic platform, Cyberette can operate as a detection layer inside existing workflows. Its partnership with Crosshub and planned integration with a European digital-forensics provider are examples of that approach.

Jakimenko positions Cyberette’s differentiation around combining detection with investigative context and evidence-ready outputs.

“We analyse how content was altered, why it was altered, and the surrounding context,” she said.

Cyberette says the platform is designed to screen media, prioritize cases and produce structured forensic evidence for human investigators.

From a friend’s photos to paying customers

Jakimenko co-founded Cyberette in 2024 after working across compliance, identity and enterprise risk and security.

She says the problem became personal when a friend’s face and body were used to create fake dating profiles that scammers used to extract money from victims.

Jakimenko built an initial prototype with a former colleague and presented it at Web Summit. Interest there led to further development, followed by funding from Rabobank and a Microsoft grant. Cyberette has since joined NVIDIA’s Inception program and the ElevenLabs startup program.

Cyberette says it has converted 85 percent of its pilots into paid contracts, a figure that has not been independently verified.

The launch follows proof-of-concept work that Cyberette says included Singapore’s Defence Science and Technology Agency, the Home Team Science and Technology Agency and forensics firm Talons Laboratories. Cyberette also says its technology is being integrated into a European digital-forensics platform that supplies police forensic units.

Child-protection organization ChildSafeNet is using the platform to verify suspected manipulated content, while Cyberette has joined the United Against Online Abuse coalition.

Seed round to fund commercial expansion

Cyberette says it now has paying customers in seven countries and is raising a seed round to expand commercial deployment. It plans to add behavioral and sentiment analysis for investigative workflows and eventually explore methods for identifying manipulation earlier in its lifecycle.

Article Topics

AI fraud  |  Cyberette  |  deepfake detection  |  deepfakes  |  funding  |  startup