Responsible AI in music: Suno’s strategy and policies

Responsible AI in music is the focus of this introduction, which examines Suno’s approach and the company’s CEO statements in response to recent concerns. The Atlantic’s AI Watchdog exposed datasets of upwards of 12 million copyrighted songs used to train AI, prompting industry discussion and company responses.

Suno says it never allows prompts for specific artists or copyrighted songs, trains using a strategy called “Original Creation, By Design”, and intentionally chose not to use artist names in training metadata. The company has said it will introduce a downloads policy designed to limit the ability to mass distribute songs on streaming platforms while preserving professional and personal ways people use Suno, and it is developing transparency tools to identify AI-generated tracks across platforms including new audio watermarking and fingerprinting technologies.

The CEO emphasized that AI should help people create something new, not imitate someone else’s work, and said the company aims to reduce the risk of unauthorised reproductions.

The Atlantic’s AI Watchdog exposed datasets of upwards of 12 million copyrighted songs used to train AI, prompting scrutiny of dataset composition and industry practices. That exposure highlighted concerns about the scale and content of training material and its relationship to existing copyrighted works. The reporting triggered responses from companies whose systems use music in training and renewed public attention on how datasets are assembled for music-generation models. The disclosure has become a focal point in discussions about accountability and dataset transparency in AI.

Suno says it never allows prompts for specific artists or copyrighted songs and intentionally chose not to use artist names in its training metadata as part of its development choices. The company describes its training approach as ‘Original Creation, By Design’, a strategy it says reduces the risk of unauthorised reproductions by prioritising new musical outputs over imitation. Suno also plans to introduce a downloads policy designed to limit the ability to mass distribute songs on streaming platforms while preserving professional, creative, and personal uses of its platform. In addition, the company is developing transparency tools to help identify AI-generated tracks across platforms, including new audio watermarking and fingerprinting technologies.

Suno enforces restrictions to reduce unauthorized reproductions by not allowing prompts for specific artists or copyrighted songs. The company intentionally omitted artist names from training metadata as part of its dataset and model-development choices. Suno describes its training approach as “Original Creation, By Design,” a strategy intended to prioritize new musical outputs and reduce the risk of unauthorized reproductions. The company’s leadership has stated that “AI should help people create something new, not imitate someone else’s work.”

Suno has said it will introduce a downloads policy designed to limit the ability to mass distribute songs on streaming platforms while preserving the professional, creative, and personal ways people use the Suno platform. The company is also developing transparency tools to identify AI-generated tracks across platforms, including new audio watermarking and fingerprinting technologies. Suno’s CEO has said the company will stay grounded in its principles and continue to listen, learn, and adapt as it navigates changes in the music ecosystem.

Mikey Shulman, the CEO of Suno, has made several statements regarding responsible AI and the company’s stance on promoting innovation over imitation in the music industry. He affirmed, “AI should help people create something new, not imitate someone else’s work,” highlighting the company’s commitment to fostering originality. Shulman also stated, “unlike many AI music platforms, we’ve never allowed prompts for specific artists or copyrighted songs,” emphasizing Suno’s policies to avoid unauthorized reproductions. Furthermore, he declared, “great music is made by people,” underscoring the value of human creativity in music creation.

Shulman outlined Suno’s future plans, noting, “We will soon introduce a new downloads policy designed to limit the ability to mass distribute songs on streaming platforms, while preserving the professional, creative, and personal ways people use Suno.” He also mentioned advancements in technology, indicating that “new audio watermarking and fingerprinting technology” would aid in transparency and authenticity verification. In a broader statement about Suno’s mission, Shulman expressed, “Staying grounded in our principles, and continuing to listen, learn, and adapt, is how we’ll navigate that change as we help build toward that future.”

Suno is introducing new transparency tools aimed at identifying AI-generated tracks, which include advanced audio watermarking and fingerprinting technologies. These technologies are designed to enhance the ability to trace the origins of music tracks generated by AI, thus promoting clarity and accountability in music production. The intention is to ensure that the music generated by AI is easily distinguishable and that any unauthorized reproductions can be quickly identified.

Through these efforts, Suno aims to maintain a responsible approach to AI in the music industry, committing to transparency and reducing the risks of copyright infringements. Audio watermarking involves embedding an indiscernible mark within the audio file, which can later be used to verify its authenticity. Fingerprinting, on the other hand, creates a unique identifier for each track, making it possible to track and monitor the use of music across various platforms. These measures are part of Suno’s broader strategy to reduce unauthorized reproductions and ensure that AI fosters new creative outputs rather than imitating existing works.

CONCLUSION

Suno presents its approach as a commitment to responsible AI in music, emphasising development choices intended to prioritise original outputs and limit imitation. The company references a training philosophy it describes as “Original Creation, By Design” and says it avoids using artist names in training metadata and permits no prompts for specific artists or copyrighted songs. Suno has announced operational measures intended to constrain large-scale redistribution on streaming platforms, including a forthcoming downloads policy, and plans to deploy transparency tools such as audio watermarking and fingerprinting to help identify AI-generated tracks.

Suno’s leadership frames these measures as part of an adaptive approach that aims to support human creativity while addressing risks around unauthorised reproductions. The CEO has stated that AI should enable people to create something new and that great music is made by people, and he has described a commitment to stay grounded in principles and to continue listening, learning, and adapting. The company says these combined steps are intended to protect creative practices and to preserve professional, creative, and personal uses of its platform as the industry evolves.