Google Lyria 3: Gemini AI Music Feature with High Privacy Security
Google Indonesia has officially introduced Lyria 3, the latest artificial intelligence-based music creation feature integrated with the Gemini ecosystem. Utilising technology from Google DeepMind, Google Lyria 3 enables users to create original songs lasting approximately 30 seconds complete with instruments, vocals, and lyrics through simple text commands.
At a media gathering in Jakarta on Friday 27 February 2026, Google Indonesia Communications Manager Feliciana Wienathan emphasised that the development of Google Lyria 3 rests on pillars of strict security and privacy. Unlike other generative AI platforms that often only consider security aspects after product launch, Google claims Lyria 3 is built on the principles of “Security by Default” and “Privacy by Design”.
Feliciana explained that user data protection is already part of the product’s initial architecture, rather than merely an additional feature. She stated: “We think about privacy as much as possible from the outset. And it’s not just something that comes up or an idea that we only think about once the product is finished.”
This is crucial given that copyright issues and voice identity abuse (deepfake) represent major challenges in the AI music industry today. The integration of Google DeepMind in Lyria 3 also brings digital watermarking technology designed to mark a work as the result of AI creation. This is done to prevent misinformation or unauthorised work claims by irresponsible parties.
Feliciana added that the model trained by Google is developed to the fullest extent to respect intellectual property rights. She noted: “The models we train are based on AI principles as much as possible. So similarity and likeness are avoided to the greatest extent possible.”
With this launch, Google reinforces its position in the global generative AI competition while ensuring that technological innovation in Indonesia upholds digital ethics and consumer protection. Lyria 3 is Google DeepMind’s most advanced AI model specifically designed to create music, vocals, and lyrics automatically through integration with Gemini AI. Users can currently generate music clips or songs lasting approximately 30 seconds. The platform is trained on AI principles emphasising originality and equipped with watermarking technology to differentiate between human and AI work.