Zoom is about to launch the much-anticipated AI-powered digital avatar feature, which will revolutionize the way users communicate asynchronously. This feature allows users to create realistic digital avatars and generate lip-synced audio with the avatar by inputting text, improving communication efficiency. However, the emergence of this technology has also raised concerns about the risks of deepfakes. The editor of Downcodes will take you to have an in-depth understanding of this new feature of Zoom, analyze its potential risks, and the security measures Zoom has taken to deal with these risks.
Zoom is gearing up to bring an amazing new feature to its users, but we might have to wait until next year to get a glimpse of it. The video conferencing giant announced at its annual developer conference that it will launch a feature that will transform user videos into lifelike AI-powered digital avatars.
This digital avatar includes not only the head, but also the upper arms and shoulders, making it a digital clone. Users simply enter what they want their digital avatar to say, and Zoom generates audio that is lip-synced to the avatar. Sounds amazing, right?
Smita Hashim, Zoom’s chief product officer, said the feature is designed to help people communicate asynchronously with colleagues in a faster and more efficient way. She emphasized that this kind of avatar can save users valuable time and energy while improving the efficiency of video creation.
However, the emergence of this technology has also raised some concerns. We have to think about whether this will bring the risk of deepfake?
In fact, many companies have developed similar AI technology that can digitally clone a person's face and match it with a fairly natural synthesized voice. For example, Tavus helps brands create avatars for personalized video ads, while Microsoft last year launched a service that can generate convincing digital avatars.
But these tools often have strict protections in place to prevent abuse. Tavus requires a verbal consent statement, and Microsoft requires its customers to obtain written permission and consent from all involved avatar talent.
By comparison, Zoom's description of its security measures seems vague. Hashim said the company is building multiple safeguards for the custom avatar feature, including advanced authentication and watermarking. She promised to continue reviewing and add security measures as needed in the future.
This move by Zoom coincides with CEO Yuan Zheng’s grand vision. In the future, he hopes to create AI that can participate in Zoom meetings, respond to emails, and even answer phone calls on your behalf.
However, the technology's emergence comes at a time when deepfakes are spreading like wildfire on social media, making it increasingly difficult to distinguish truth from disinformation. Deepfake videos involving President Joe Biden, Taylor Swift and Vice President Kamala Harris have received millions of views and shares online this year. Recently, in the wake of Hurricane Helene, a flood of fake AI-generated images flooded the web, showing fictional scenes of destruction and human suffering.
Deepfake technology has also been used to target individuals, such as posing as relatives to commit fraud. According to the Federal Trade Commission, losses related to impersonation scams exceeded $1 billion last year.
So how exactly does Zoom prevent scammers from using its tools to generate malicious fake videos? It's unclear. A mockup provided by the company shows a visible watermark in the upper right corner of the custom avatar video. But this watermark is easily cropped by screen recording tools.
We hope to know more details when Zoom plans to release this feature in the first half of 2025. At that point, users will have to pay an additional $12 per month to use custom avatars in Zoom Clips, its asynchronous video tool.
Regardless of what measures Zoom ultimately takes, regulatory efforts are underway to try to curb the proliferation of deepfakes. In the absence of laws criminalizing deepfakes at the federal level, more than 10 states have enacted regulations targeting AI-assisted impersonation. California's law, which is currently on hold, would be the first to give judges the power to order publishers of deepfake content to remove the content or face possible fines.
As Zoom launches this new feature, we can’t help but ask: Does technological progress always have a positive impact? While enjoying convenience, have we inadvertently opened Pandora’s Box? This requires each of us to think deeply. .
Although Zoom’s AI digital avatar feature is promising, it also has potential risks. How to strike a balance between technological progress and security risks will be a question that Zoom and the entire technology industry need to seriously consider in the future. The editor of Downcodes will continue to pay attention to the progress of this technology and bring you more related reports.