Scientists at IDEAS NCBR, a Polish AI-focused research center, have developed a prototype program designed to verify whether an online conversation was conducted with a real person.
Deepfake technology, which uses artificial intelligence to create realistic but fake multimedia content, is increasingly used for disinformation, manipulation, and fraud, despite its legitimate applications in entertainment and education.
Recent instances in Poland, such as fake videos of President Andrzej Duda, footballer Robert Lewandowski, and Rafał Brzoska, CEO of postal delivery service InPost, promoting fraudulent ‘investment programs,’ have revealed the growing threat.
“Deepfakes are becoming more sophisticated, making detection increasingly challenging and posing new digital security challenges,” said researchers from IDEAS NCBR.
Their experimental project, ProvenView, was developed by postdoctoral researcher Shahriar Ebrahimi and Ph.D. student Parisa Hassanizadeh.
Working in the Systems Security and Data Privacy group led by Prof. Dr. Stefan Dziembowski, researchers have utilized Zero-Knowledge Proof (ZKP) technology—a cryptographic method that verifies the truth of information without revealing the details.
"ZKP technology was once considered purely theoretical and closer to pure mathematics than practical computer science applications. ProvenView exemplifies how ZKP can address contemporary digital threats," said Prof. Dziembowski.
ProvenView aims to authenticate online interactions and protect content. After an online meeting, users can verify if the video genuinely originated from the other party's webcam.
In a simplified process, the original video creator generates a digital proof of authenticity, confirming the material’s origin. This proof can be embedded in a video recording application or a separate program. Viewers can then use this proof to verify the video's authenticity.
“This solution can be useful both for individual users, who could use it to authorize online meetings, and for content creators, who could use it to strengthen the protection of their image published in a YouTube video against theft and use, for example, in pornography,” Ebrahimi said.
ProvenView is an experimental solution and is not yet ready for full implementation. Ebrahimi noted that authenticity proofs are generated post-video creation, taking about an hour to verify after a conversation.
However, future developments aim for real-time video authenticity verification.
ProvenView received recognition in May 2024 by winning an award at the ZK Hack Krakow hackathon.
Recent instances in Poland, such as fake videos of President Andrzej Duda, footballer Robert Lewandowski, and Rafał Brzoska, CEO of postal delivery service InPost, promoting fraudulent ‘investment programs,’ have revealed the growing threat.
“Deepfakes are becoming more sophisticated, making detection increasingly challenging and posing new digital security challenges,” said researchers from IDEAS NCBR.
Their experimental project, ProvenView, was developed by postdoctoral researcher Shahriar Ebrahimi and Ph.D. student Parisa Hassanizadeh.
Working in the Systems Security and Data Privacy group led by Prof. Dr. Stefan Dziembowski, researchers have utilized Zero-Knowledge Proof (ZKP) technology—a cryptographic method that verifies the truth of information without revealing the details.
"ZKP technology was once considered purely theoretical and closer to pure mathematics than practical computer science applications. ProvenView exemplifies how ZKP can address contemporary digital threats," said Prof. Dziembowski.
ProvenView aims to authenticate online interactions and protect content. After an online meeting, users can verify if the video genuinely originated from the other party's webcam.
In a simplified process, the original video creator generates a digital proof of authenticity, confirming the material’s origin. This proof can be embedded in a video recording application or a separate program. Viewers can then use this proof to verify the video's authenticity.
“This solution can be useful both for individual users, who could use it to authorize online meetings, and for content creators, who could use it to strengthen the protection of their image published in a YouTube video against theft and use, for example, in pornography,” Ebrahimi said.
ProvenView is an experimental solution and is not yet ready for full implementation. Ebrahimi noted that authenticity proofs are generated post-video creation, taking about an hour to verify after a conversation.
However, future developments aim for real-time video authenticity verification.
ProvenView received recognition in May 2024 by winning an award at the ZK Hack Krakow hackathon.
Source: TVP World, PAP Nauka
More In Society MORE...
