Cyberette is building forensic-grade AI for a post-truth internet

Cyberette, a Dutch startup founded in 2024, is developing AI-powered software designed to detect and analyze manipulated digital content, including deepfakes and altered media. This technology addresses the growing problem of sophisticated digital manipulation, aiming to provide solutions for fraud detection and investigations. The company's work is particularly relevant as the cost of these deceptions increasingly falls on victims, rather than the platforms hosting the content. The Cyberette platform employs several techniques, including landmark-based detection to identify facial inconsistencies and heatmap-based analysis to highlight altered areas. The system also performs sentiment analysis and offers real-time detection, delivering results in under two seconds. These features are complemented by watermarking and metadata analysis, with a focus on providing provenance information to support deeper investigations by governments, media organizations, and enterprises. Cyberette's technology has a significant impact on investigative workflows, especially regarding deepfakes and manipulated content. The platform's accuracy rate of 99.7% across tested datasets and its real-time, low-latency performance at scale enable faster, more reliable content authentication. This capability is crucial for supporting investigative teams by providing explainability and structured evidence in the face of increasingly complex digital manipulation. Cyberette's approach is designed to provide actionable intelligence for investigators. The focus on explainability, along with detailed analysis of manipulation patterns and potential sources, sets it apart from other tools. With a strong backend architecture and lightweight models, Cyberette offers cloud-based or on-premise deployment, accommodating various customer infrastructure preferences.