Avatarify Deepfake App to Add Digital Watermark

The Rise of Mobile Deepfakes with Avatarify

The creation of convincing deepfake videos, once a complex undertaking, has become remarkably accessible. Currently, a smartphone is often all that’s required. Avatarify, a company enabling deepfake video creation directly on mobile devices – bypassing cloud processing – has experienced rapid growth in app store rankings, gaining traction even among prominent figures like Victoria Beckham.

Addressing the Issue of Authenticity

A significant concern surrounding many deepfake videos is the absence of a clear indicator confirming manipulation. To combat this, Avatarify has announced plans to implement a digital watermark system, designed to help identify altered video content.

Company Overview and Growth

Though operating from Moscow with a U.S. headquarters, Avatarify was initially launched in July 2020. Since then, the application has been downloaded millions of times. The company reports that 140 million deepfake videos were generated using their platform in the current year alone.

On TikTok, videos utilizing the hashtag #avatarify have amassed 125 million views. While facing competition from established players like Reface, Snapchat, Wombo.ai, Mug Life, and Xpression, Avatarify has achieved this success without substantial venture capital funding, relying on an initial angel investment.

Bootstrapped Success and Revenue Model

Despite securing only $120,000 in angel funding, the company has deliberately avoided further venture capital investment. Avatarify claims to have organically grown to nearly 10 million downloads and is achieving an estimated $10 million annual revenue, all with a team of fewer than 10 individuals.

This is largely attributable to their freemium subscription model. Users are offered a 7-day free trial, followed by subscription options of $34.99 for a 12-month period or $2.49 per week. The core app features remain available without a subscription, but generated videos will display a visible watermark.

Privacy Considerations

The founders emphasize the app’s commitment to user privacy. Video processing occurs directly on the user’s device, rather than in the cloud, mitigating the risk of data breaches.

How Avatarify Works

Avatarify utilizes machine learning algorithms to animate faces from user-provided photos, creating short videos with added sound. Users select an image, choose desired effects and music, and then initiate the animation process. These videos can then be shared on platforms like Instagram or TikTok.

TikTok Popularity and Ease of Use

The app’s growing popularity on TikTok is fueled by its simplicity; users no longer need to master complex choreography or demonstrate significant creativity – simply animating a celebrity photograph is sufficient.

Avatarify’s creators state that the app is not intended for impersonation, although enforcing this restriction presents a considerable challenge.

Origins and Viral Launch

Co-founders Ali Aliev and Karim Iskakov developed the application during the COVID-19 lockdown in April 2020. Ali Aliev initially spent two hours coding a Python program to transfer his facial expressions onto another person’s face within a Zoom filter. The resulting real-time video stream, featuring Elon Musk’s face, astonished participants on the call.

The team subsequently shared the video, which quickly went viral. The code was published on Github, leading to a surge in downloads. By March 19, 2021, the repository had been downloaded 50,000 times.

From Samsung to App Store Success

Ali Aliev left his position at the Samsung AI Centre to focus on the app’s development. Following the release of Avatarify’s iOS app on June 28, 2020, viral videos on TikTok propelled it to the top of the App Store charts without any paid advertising. In February 2021, Avatarify was ranked as the top free app globally.

Between February and March 2021, the application generated over $1 million in revenue (Source: AppMagic).

Ongoing Concerns Regarding Deepfakes

Despite Avatarify’s achievements, the ethical concerns surrounding deepfake technology persist, including the potential for misuse in creating nonconsensual pornography and the unauthorized use of individuals’ likenesses.