Paravision brings deepfake detection to the market after new deal with Five Eyes country
U.S. computer vision developer Paravision announced its plans to develop a product from its prototype deepfake detection platform and bring it to the market after securing another contract from an unnamed country that is part of the Five Eyes security alliance.
The additional contract comes after the company clinched its first deal with the “major” Five Eyes government in June last year. The alliance comprises security agencies from the U.S., Australia, Canada, New Zealand and the UK.
Paravision’s platform includes an AI-based deepfake detector and a web-based, on-premises application that allows human experts to review suspected synthetic media and confirm whether they are genuine or fake, the company explains in its release. The program expansion covers a period of one year.
The deal represents a continuation of cooperation with government agencies for the San Francisco-based company.
This month, Paravision’s president and chief operating officer Benji Hutchinson was appointed to the U.S. Department of Commerce National Artificial Intelligence Advisory Committee (NAIC) which provides recommendations on several topics related to AI competitiveness. The company has also been working with the U.S. Air Force on a biometric identity solution since 2020.
Within the past year, global governments and tech companies around the world have been taking steps to combat synthetic media, which includes a range of different data, imaging scope and types of manipulation.
“Synthetic imagery in general and deepfakes specifically pose substantial risks to identity, privacy, democracy, national security, and human rights,” says Joey Pritikin, Paravision chief product officer. “We are proud to continue our work with a respected government partner and to have their support as we bring advanced deepfake detection technology to market.”
Paravision’s deepfake detection is powered by its facial recognition tech which has achieved success in accuracy benchmarks such as NIST FRVT and DHS Biometric Technology Rally. The deepfake detection AI engine has already demonstrated accuracy on par with the best-in-class deepfake detectors in academic literature, the company states in the release.