Detecting and Preventing
Harmful and Illegal Content in Images & Video

SafeToWatch is a ground-breaking real-time image & video threat detection technology. SafeToWatch is able to detect harmful content such as child sexual abuse material (CSAM), pornography, suggestive imagery and hentai (cartoon pornography) in real-time.
SafeToWatch is the only solution in the world to be able to detect even uncategorized CSAM in real-time.
SafeToWatch moderates content on your application, network or platform without human intervention and helps you get ahead of your competitors with upcoming child safety legislation.

How Does It Work?
SafeToWatch automatically determines whether visual data represents undesirable and illegal content such as pornography, sexually suggestive imagery, and cartoon pornography. It stops content from being viewed, distributed, or created once undesirable or illegal content is detected using machine learning (ML).
SafeToWatch has the unique ability to detect both already categorised and uncategorised first-generation Child Sexual Abuse Material (CSAM) by leveraging state of the art machine learning techniques. This unparalleled technology allows SafeToWatch to prevent the most severe child abuse being consumed and distributed online.
SafeToWatch has been developed in collaboration with the Internet Watch Foundation.
Implementation & availability
SafeToWatch is a Software Development Kit for the analysis of image and video data. As an SDK, SafeToWatch can be embedded at any level: application, system, or network.
At the application level: App vendors can ensure that the content captured from the camera by their application is always safe, or that the images transmitted are filtered.
At the system level, on fully supervised mobile devices, a device management software can embed the SafeToWatch SDK and monitor the camera preview for any application trying to use the camera. For manufactured Android devices, SafeToWatch can be embedded into the system image to always filter the camera transparently for all applications.
Finally, on the network, SafeToWatch can be added to a network data inspection pipeline to process streamed images and videos, identifying harmful content before they reach the device.

