New web service allows users to create their own deep fake videos

You've probably heard about so-called 'deep fake' videos, which are digitally altered to make people look, or sound, like someone else. 

Until now, the necessary technology has been out of reach to most, but a new web service intended as a sophisticated business tool, offers to bring it right to your computer.


Synthesia is a new London-based web service that offers artificial-intelligence video creation for training, how-to, and marketing presentations. Houston tech-expert Juan Guevara Torres worked with an early version of the service. 

"You know it's fake, but your eyes are telling you it's not."

Starting with dozens of models, or avatars, as they're called, users select from a variety of languages and accents to load a script and process a video that shows the avatar speaking naturally. For businesses, it's a potential money-saver. 

"They're removing the need for a professional talent for these types of videos," says Guevara Torres.


For people worried about how the occasional deep fake video can fake-out the casual observer, the easy ability for anyone to create real-looking people to say whatever they're programmed to say may be reason to pause. 

On the internet, and social media, there is already a willingness for people to believe what they see and hear, and critics warn these artificial-intelligence videos could cause real harm if consumers aren't critical. 

"I think we have this responsibility, before we disseminate these things, to really make sure what we're sharing is accurate," says University of Houston Professor Trevor Boffone.


Synthesia says it runs all the content, created through its website, through an internal screening process to ensure it's not sharing any ugly messages. 

Meantime, as remarkable as these initial versions are, they still have enough flaws that most people would recognize it's not real. Later versions, however, will be harder to tell, leaving consumers with a greater responsibility to be skeptical.