Subscribe
Sign up to stay up to date with umanitek's latest developments.
Sign up to stay up to date with umanitek's latest developments.
Zug, Switzerland/Reykjavik, Iceland. August 20th, 2025.
Umanitek, a Swiss AI company focused on reducing harm from AI, and Videntifier, a global leader in forensic visual content identification technologies, today announced a strategic partnership aimed at removing illegal content from the internet, such as deepfakes and child explicit sexual material (CSAM).
In a major milestone, Videntifier and Umanitek have jointly created the largest “immune system” on the planet, designed for identifying and removing illicit content on the internet.
The partnership brought together Umanitek’s Guardian Agent network – decentralized collection of labelled content built on blockchain-based OriginTrail network, with Videntifier’s ability to recognize the most extensive set of known illegal visual content. Together they enable checking against the world’s largest shared library, containing over millions of hours of videos and tens of millions images of harmful content. Videntifier has already established itself as one of the leading providers of accurate, robust and fast recognition of illegal and harmful content, trusted by Interpol, NCMEC, IWF and a number of other law enforcement agencies and hotlines.
Through this partnership, for the first time, tech platforms and websites can use Umanitek’s AI Agent to query the decentralized library of content before allowing content up on their platforms. If someone has seen the content before and labelled the content illicit, illegal or copyright infringed, then platforms now have a way to prevent it from being published or redistributed.
Such a system has not been accessible until today. Most tech platforms must avoid sharing raw data with each other, due to privacy, legal and competitive constraints. Umanitek and Videntifier collectively have created a way to solve it: they haven’t just enabled access to the world’s biggest repository of content that should not be further distributed, but more importantly created a quick, cost-effective and privacy-preserving way for tech platforms to check ownership and legality of content before uploading/distributing.
The applied technical architecture based on OriginTrail Decentralized Knowledge Graph ensures that every content owner keeps their data in their own environment, never sharing the underlying raw files (i.e. a video) with anyone else. Instead, all data stays within each participant’s own infrastructure. Uniquely, Videntifier and Umanitek allow the network participants to create a digital fingerprint of the raw data using Videntifier’s forensic image and video recognition and fingerprinting technology. The encryption does not only uniquely encrypt a piece of content but, importantly, allows for fingerprints to be compared between each other for likeness. This allows everyone to check and compare a piece of content against the entire decentralized library of fingerprints, without ever exposing the underlying video or photo. In short, the solution allows platforms to share and compare data without actually giving up the underlying data. This is a game changer.
By saving massive amounts of time and costs in analyzing confiscated laptops or cloud folders of suspected CSAM. With this system, law enforcement officers can match suspect files to known illicit content far more efficiently, moving much faster towards prosecution and providing more rapid relief to victims. Law enforcement and government agencies already operating on Umanitek’s network can now both communicate unwanted content/files, and compare flagged or illegal files without needing to share the underlying raw files.
The integration between Umanitek and Videntifies creates a powerful digital immune system for the internet. By combining Videntifier’s high-speed, forensic-grade digital content matching engine with Umanitek’s AI-enabled reasoning capabilities over a decentralized knowledge graph based on OriginTrail, we have created a cost effective way to find and eliminate unwanted content on the internet, such as deep fakes, CSAM and through that also possibly limit the AI hallucination problem.
“In Videntifier, we’ve found a partner whose forensic precision and institutional trust make them an ideal addition to the Umanitek AI Guardian network,” said Chris Rynning, Chairman of Umanitek. “Videntifier and Umanitek have created one of the largest databases of illegal content in the world combined with an AI agent that actively looks for and takes down illegal or unwanted content globally”.
“Joining the Umanitek Guardian Network allows our fingerprinting capabilities to be applied within an ecosystem that prioritizes trust, transparency, and real-world impact,” said Ari Jonsson, CEO of Videntifier. “Together, we are creating AI agents that enable platforms and institutions to meet the urgent challenges of fighting harmful content online.”
The partnership has already passed the initial technical integration phase, with the joint solutions expected to launch in October this year. Journalists, internet platforms, law enforcement agencies, non-governmental organizations, and other stakeholders looking to reduce the appearance of illegal content online are encouraged to contact the companies via umanitek.ai or videntifier.com to schedule a demo already now!