- New deepfake detection tool helps tackle fake content
- A “deepfake score” helps users recognize AI-generated video and audio
- The tool is free to use to help limit the impact of fake content
To combat the increase in such scams, security company CloudSEK has launched a new one Deep fake detection technologydesigned to counter the threat of deepfakes and provide users with a way to identify manipulated content.
CloudSEK’s detection tool aims to help organizations identify deepfake content and prevent potential damage to their operations and credibility by assessing the authenticity of video frames, focusing on facial features and motion inconsistencies that could indicate deepfake tampering, such as facial expressions with unnatural transitions and unusual textures in the background and on faces.
The rise of deepfakes, but there is a solution
Audio analysis is also used, where the tool detects synthetic speech patterns that signal the presence of artificially generated voices. The system also transcribes audio and summarizes key points, allowing users to quickly assess the credibility of the content they are reviewing. The end result is an overall ‘Fakeness Score’, which indicates the likelihood that the content has been artificially altered.
This score helps users understand the level of potential manipulation and provides insight into whether the content is AI-generated, mixed with deepfake elements, or likely human-generated.
A Fakeness score of 70% and above is AI-generated, 40% to 70% is questionable and possibly a mix of original and deep-fake elements, while 40% and below is likely human-generated.
In the financial sector, deepfakes are used for fraudulent activities such as manipulating stock prices or misleading customers with fake video-based KYC processes.
The healthcare sector has also been hit, with deepfakes being used to create fake medical records or impersonate doctors, while government agencies are threatened by election-related deepfakes or falsified evidence.
Likewise, the media and IT sectors are equally vulnerable, with deepfakes being used to create fake news or damage the reputation of brands.
“Our mission to predict and prevent cyber threats extends beyond businesses. That’s why we decided to release the Deepfakes Analyzer to the community,” said Bofin Babu, co-founder of CloudSEK.
