HeyGen and AI Voice Translation: the future of language translation or another privacy threat?
Image: ‘Machine Learning and Artificial Intelligence’ by Mike MacKenzie is licensed under CC BY 2.0 DEED
Generative AI platform HeyGen shapes the future of voice translation and video production, forcing us to redefine the concept of privacy and data protection.
Founded in 2020, the generative AI platform HeyGen enables users to create studio-quality videos, add voice narration, and automatic translations. No previous skill or knowledge is required; all customers need to do is sign up. The platform provides a selection of hyper-realistic avatars, pre-recorded voices, and video templates that can easily be customised. However, convenience and efficiency come at a price, as using human images created by generative AI causes concerns about privacy and data protection. More importantly, it forces us to reflect on how we understand privacy in a world where anyone can create a deep fake from the comfort of their sofa.
After signing up, a customer can choose a provided video template, add one of the 120 hyper-realistic avatars in either a studio or natural setting, and pair it with one of 300 distinct voices provided in over 40 languages. Once the video is made, it can be instantly translated into any available language. Possibilities are truly endless, given that if the selection is not satisfying users can create their avatars and record their real-life voice, which can be later shared with the whole community. Later, others can use the voice and avatars in their projects without crediting the author. This novel generative AI tool transports video-making to a new level, redefining the meaning of personalized and tailored content.
HeyGen not only enables effortless video-making, but it also transforms the world of translations. Along with similar technologies, generated text can be almost instantly translated into a significant number of languages. Generative AI can additionally incorporate cultural and contextual cues, which previously was not possible with simpler AI such as Google Translate. Now, previously troublesome language barriers can be easily broken, allowing us to communicate with almost anyone in the world. On the one hand, such advancements could make people’s personal lives a lot easier. On the other, the commercial use of personal images and voices by AI could cause privacy concerns.
Although visiting the HeyGen page users are provided with a detailed ethics and privacy policy, the alarming growth of deepfakes can justifiably cause privacy concerns. Following regional GDPR, the company introduced a “mix of human and AI oversight” to ensure the safety of its customers. To establish trust, the HeyGen team highlights strict ethical procedures, transparency, control, and making the content private as a default. At the same time, ensuring that the content will never be used for identification. However, with users' consent, the personal avatars can be further used for AI training. Therefore, by prioritizing efficient video making, customers are making a leap of faith, trusting the company to protect their image.
Strict laws and privacy policies can prevent users from misusing the platform and developing deep fakes in malicious manners. However, allowing the construction of new personalities through voice, expressions, and ideas presented by the avatars, the technology forces us to reflect on our definitions of privacy. Are data protection laws and regulations enough? To what extent do we feel comfortable with sharing our image to be redefined? Although AI opens new possibilities in video-making and translation, it also challenges the ideas of privacy and self-expression.