Microsoft’s New Tool Raises Concerns Over Deepfake Ethics
Microsoft has unveiled a new tool at the Ignite 2023 event that allows users to create photorealistic deepfake avatars. The Azure AI Speech text to speech avatar feature enables users to generate videos of an avatar speaking by uploading images of a person and writing a script. A model trained by Microsoft’s tool drives the animation, while a separate text-to-speech model reads the script aloud. This tool has various applications, including creating training videos, product introductions, customer testimonials, conversational agents, virtual assistants, and chatbots.
Ethical Concerns Surrounding Deepfake Avatars
While the new tool opens doors for innovation, there are ethical concerns surrounding the potential misuse of deepfake technology. Similar avatar-generating tech from AI startup Synthesia has been exploited to create propaganda in Venezuela and spread false news promoted by pro-China social media accounts. As a precautionary measure, Microsoft limits access to custom avatars and requires registration for certain use cases. However, the company remains silent on whether companies should compensate actors for their AI-generated likenesses or label avatars as AI-generated.
Microsoft’s Personal Voice Tool
In addition to the deepfake avatars, Microsoft has also introduced the personal voice tool within its custom neural voice service. This feature allows users to replicate their own voice in a matter of seconds using a one-minute speech sample. Microsoft promotes personal voice as a way to create personalized voice assistants, dub content into different languages, and generate bespoke narrations for stories, audiobooks, and podcasts. To avoid potential legal issues, Microsoft requires users to provide explicit consent before synthesizing their voices and restricts the usage of personal voice to specific applications.
Compensation and Identification Challenges
Microsoft has not addressed the issue of compensating actors for their contributions to the personal voice tool. It remains to be seen how the company plans to navigate this potential legal minefield. Additionally, there are concerns about identifying AI-generated voices. TechCrunch questioned whether Microsoft intends to implement watermarking technology to make AI-generated voices more easily identifiable, but the company did not provide a response.
While these tools showcase the potential of AI technology, they also highlight the need for proper regulations and ethical considerations to prevent the misuse of deepfake avatars and protect the rights of individuals involved.