Entrust’s 2027 Identity Fraud Report, released today, found that deepfake detections rose 165% year over year, and 462% in Europe. The report highlights virtual camera injection, where criminals use software to feed fake or stolen video directly into identity verification tools. The “live” selfie the system checks never comes from a real camera. Rohan Gupta, Co-founder and CEO of Diopter, which detects deepfakes and AI social engineering on live video and voice calls, previously founded QuillBot and grew it to more than 35 million monthly users. Here’s his take:
Rohan Gupta, CEO and Co-Founder, Diopter AI:
“A 165% jump in deepfake detections in one year, and 462% in Europe, tells you how fast deepfake fraud is scaling. The virtual camera technique in Entrust’s report stands out because it skips the camera entirely. The verification software thinks it’s watching a live person when it’s receiving whatever video the fraudster chooses, synthetic or stolen. The same virtual camera tools work on Zoom and Teams calls, which is where fraudsters impersonate executives to push wire transfers, vendors to change bank details, and job candidates to get hired. Checking a single frame for signs of manipulation won’t hold up as the fakes keep improving. Security teams also need to look at what surrounds the video, including the device and session it’s coming from and how the conversation unfolds, especially when someone suddenly adds urgency, asks for secrecy, or makes an unusual request.”
This is bad. And if I may be so bold, action needs to be taken to stop this attack vector. ASAP.
Related
This entry was posted on November 18, 2026 at 12:56 pm and is filed under Commentary with tags Entrust. You can follow any responses to this entry through the RSS 2.0 feed.
You can leave a response, or trackback from your own site.
Deepfake detections jumped 165% in a year as fraudsters feed fake video straight into identity checks
Entrust’s 2027 Identity Fraud Report, released today, found that deepfake detections rose 165% year over year, and 462% in Europe. The report highlights virtual camera injection, where criminals use software to feed fake or stolen video directly into identity verification tools. The “live” selfie the system checks never comes from a real camera. Rohan Gupta, Co-founder and CEO of Diopter, which detects deepfakes and AI social engineering on live video and voice calls, previously founded QuillBot and grew it to more than 35 million monthly users. Here’s his take:
Rohan Gupta, CEO and Co-Founder, Diopter AI:
“A 165% jump in deepfake detections in one year, and 462% in Europe, tells you how fast deepfake fraud is scaling. The virtual camera technique in Entrust’s report stands out because it skips the camera entirely. The verification software thinks it’s watching a live person when it’s receiving whatever video the fraudster chooses, synthetic or stolen. The same virtual camera tools work on Zoom and Teams calls, which is where fraudsters impersonate executives to push wire transfers, vendors to change bank details, and job candidates to get hired. Checking a single frame for signs of manipulation won’t hold up as the fakes keep improving. Security teams also need to look at what surrounds the video, including the device and session it’s coming from and how the conversation unfolds, especially when someone suddenly adds urgency, asks for secrecy, or makes an unusual request.”
This is bad. And if I may be so bold, action needs to be taken to stop this attack vector. ASAP.
Share this:
Like this:
Related
This entry was posted on November 18, 2026 at 12:56 pm and is filed under Commentary with tags Entrust. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.