Synthesia, an Nvidia-backed AI company, introduced a new set of AI-powered avatars on Thursday that can express human emotions based on text inputs from users.

Synthesia's Expressive Avatars

London-based company Synthesia displayed its "expressive avatars" in a video shared on the social media platform X on Thursday, demonstrating various emotions such as happiness, sadness, and frustration. This technology arrives amid growing concerns about deepfake videos, which can convincingly impersonate individuals for scams or misinformation purposes.

Synthesia's avatars can blur the boundary between virtual and real characters to streamline professional video production by removing the need for cameras, microphones, actors, extensive edits, and associated costs. Synthesia operates a studio in London, where actors read scripts in front of a green screen to train the system.

In a demonstration, the company displayed three lines of text on its platform: "I am happy. I am sad. I am frustrated." The AI-generated actor in the video then responded by reading the text in the tone reflecting each corresponding emotion.

The avatars are trained using the movements of real-life actors who read scripts in front of a green screen, which Amazon Web Services showcased on Wednesday, with an avatar version of AWS' managing director Tanuja Randery. 

READ ALSO: Paradox.Ai: Companies Face Growing Controversy as Hiring Mandates Personality Assessments with Blue Humanoid Aliens 

Digital Avatars in Corporations

Synthesia's technology is being utilized by almost 55,000 businesses in creating digital avatars for corporate presentations and training videos.

Founded in 2017 with a focus on developing dubbing and lip-synching tools, the company shifted its focus to launch its first AI avatars tailored for corporate clients in 2020. This strategic move paid off, as last year Synthesia achieved unicorn status, reaching a valuation of over €1 billion, which places it among Britain's newest AI unicorn companies. Shareholders include Accel, Kleiner Perkins, GV, FirstMark Capital, and MMC.

Potential Misuses and Preventions of Such Avatars

Synthesia responded to worries about the potential misuse of its videos for generating fake news content by requiring publishers to register as enterprise customers to access synthetic avatars. While content produced using its technology undergoes scrutiny by moderators, the company does not publicly reveal pricing details for its enterprise customers.

The company also mandates that all new clients undergo a comprehensive Know Your Customer process, similar to those employed by the banking sector to help deter malicious actors from fabricating fake company profiles to disseminate misinformation.

Synthesia has affirmed its readiness for the upcoming global elections and has put in place various controls to prevent its platform from being exploited by hostile actors aiming to manipulate the outcome of different votes.

The company is also a member of the Coalition for Content Provenance and Authenticity, a group of AI firms dedicated to implementing content credentials and digital watermarking for AI-generated content, which ensures that viewers can discern whether the content is created by artificial intelligence rather than by a human.

RELATED ARTICLE: Bard to Gemini: Google CEO Acknowledges "Problematic" Mistakes, Pulls AI Features Offline for Further Testing