Microsoft Unveils Voice Cloning Feature, Raising Urgent Security Questions

Microsoft is introducing a new feature in Teams that will allow users to clone their voices for real-time interpretation in multiple languages. 

Announced at Microsoft Ignite 2024, the “Interpreter” tool will let users simulate their own voice while speaking in up to nine languages, including English, Spanish, Mandarin, and French, starting in early 2025.

The voice simulation feature is designed to offer a more personal, engaging experience during meetings, allowing users to speak in different languages while retaining their voice’s unique tone and cadence, according to Tech Crunch. 

Microsoft emphasized that the tool will only be available to Microsoft 365 subscribers, and that the voice cloning feature requires user consent, which can be disabled through settings. 

The company also confirmed that no biometric data is stored, and the voice simulation will remain faithful to the original speaker’s message without adding assumptions or extra information.

While the new tool promises to streamline multilingual communication in global teams, it raises concerns about security and misuse. 

AI-driven voice cloning has been used to create deepfakes, posing risks in areas such as fraud and disinformation. Microsoft’s feature is intended for business use, but its potential for abuse, including in phishing scams, remains a concern.

Despite the risks, Microsoft’s new tool reflects the growing trend in AI translation technology, which is expected to become a $35.1 billion market by 2026. 

Competitors like Meta and ElevenLabs have also rolled out similar technologies, although AI translations are still often less nuanced than human interpreters, particularly in conveying colloquialisms and cultural context.

Microsoft has yet to detail the full safeguards for the Interpreter feature, leaving questions about its potential vulnerabilities in the hands of malicious actors.