Microsoft (MSFT) and Adobe (ADBE) are diving into the battle against deepfake technology, with a $26 million investment into image verification startup Truepic, pointing to the growing threat brought on by digitally altered photos and videos.
Speaking to Yahoo Finance Live, Samir Kumar, managing partner of Microsoft’s Venture arm M12, said image verification technology developed by the San Diego-based firm would be a "critical part of the defense against misinformation," adding that reliance on forensic techniques driven by artificial intelligence technology alone is a "losing battle."
“The fundamental AI approach that’s being used to generate these fakes has built into it the ability to get better over time and learn from when we find new forensic techniques to capture these fakes, which will ultimately lead to better fakes,” Kumar said. “So you can't rely on forensic approaches to do this. You have to be able to say, we have to trust where this content came from. Was it captured in specific conditions that are still true?”
Founded in 2015, Truepic uses controlled capture technology that locks in a photo or video the instant it is created. That allows users to authenticate the “provenance” data of an image and certify the time, date, and location at the moment of capture.
The latest funding round led by M12 includes Sony Innovation Fund by IGV, Hearst Ventures and investors from Stone Point Capital. It brings the total amount raised to $36 million for the startup.
Truepic CEO Jeffrey McGregor said the company’s patented technology has verified millions of photos and videos captured in more than 150 countries so far.
“There's just been a secular trend of disinformation, fraud, deception, and it's really affected every area of the internet, including the services we use every day, including social media, including the political environment,” said Truepic CEO Jeffrey McGregor. “Being able to restore trust in the visual media that we look at as consumers every day can have a profoundly positive impact on the internet and society.”
“Deepfakes” represent audio and video content manipulated by artificial intelligence. The technology, once limited to Hollywood special effects studios, has proliferated at a rapid rate in recent years, as online tutorials and forums increasingly democratize AI tools.
'A very real problem that we can solve'
McGregor said the scope of its application has also evolved beyond content on popular social media platforms and political websites.
“The same tools are being used to defraud businesses,” McGregor said. “So synthetic media can be created to generate fake head shots or damage on a vehicle and enterprise is really left in a difficult position right now, because they don't have tools that can help them determine authenticity.”
That has led to key partnerships with more than 100 enterprise customers including Ford and Equifax, as well as the United National Capital Development Fund (UNCDF). Last year, Truepic forged a partnership with Qualcomm (QCOM), to integrate its control capture capture technology into Qualcomm-powered phones, and images and videos captured by the devices.
Adobe, a key backer in the funding round, is working to integrate Truepic technology across its Creative Cloud Suite. Together, the firms have developed an open standards body, to make it easier for users to access the metadata of any given image, and verify and trace the way in which that image has been edited.
“We're giving you sort of the tools and the receipts,” said Will Allen, vice president of product at Adobe. “We're not under the illusion that this thing's going to get out into the world and magically misinformation is going to disappear. It's as old as human beings. But we think that's a very real problem that we can solve, if we amplify the good actors that are out there.”
Akiko Fujita is an anchor and reporter for Yahoo Finance. Follow her on Twitter @AkikoFujita