dot, the Deepfake Offensive Toolkit from Sensity AI, creates real-time, controllable deepfakes that can be injected into a virtual camera feed. It was built for offensive security work: penetration testing identity verification and video conferencing systems, carried out by security analysts, red teams and biometrics researchers.
The toolkit takes your webcam feed, applies a deepfake transformation, and outputs the result through a virtual camera so it appears as a live participant in Zoom or other conferencing software. Its methods need no additional training โ a single photo becomes the impersonation target on the fly. Supported techniques include face swapping via SimSwap at multiple resolutions with optional GPen face super-resolution, a lighter OpenCV-based face swap, and image animation via the First Order Motion Model (FOMM). dot can be run through a graphical interface or a command line, inside Docker, against pre-recorded image and video files, or even through an Android emulator, and it runs on GPU, Apple Silicon or (slowly) CPU. The project is BSD-3-Clause licensed, has more than 4,500 stars on GitHub, and its README states clearly that it is developed for research and demonstration, with users responsible for complying with applicable laws.
Free. dot is open source under the BSD-3-Clause license and runs on your own hardware.
Red teams, security analysts and biometrics researchers who need to test how identity verification, liveness checks and conferencing systems hold up against real-time deepfake impersonation, in authorized engagements.