Suno positions itself as a reference platform for generating music with artificial intelligence, with a clear policy that prohibits the use of copyrighted material as seed. However, a recent investigation demonstrates that its filtering systems have evident flaws. The protection can be bypassed with simple methods and commonly available tools, which questions the real effectiveness of these controls in such a sensitive creative environment.
Audio obfuscation technique to evade detection 🕵️
The process to bypass Suno's filters does not require advanced knowledge. It consists of taking a protected MP3 file and processing it with free software like Audacity. Slightly altering the speed or pitch of the track, and adding a subtle layer of white noise, modifies the file's digital fingerprint. These changes are minimal to the human ear, but sufficient to confuse detection algorithms. Thus, the platform accepts the audio as an original seed, allowing the generation of covers or new derivative pieces from copyrighted works.
The renaissance of the mixtape, now with an AI seal 🎵
It seems that the digital era has found its equivalent to the trick of recording songs from the radio with a cassette, but with an algorithmic twist. Users, instead of waiting for the DJ to speak, now use white noise and speed changes to fool a virtual DJ that is much stricter in theory. Human creativity to share music always finds a way, even if that way involves adding static to a digital file so that an AI considers it sufficiently unique. The paradox is clear: we use advanced technology to imitate analog evasion tactics.