@AvrilNewsFans@AvrilLavigne Consent is the useful line here: training and generation are different questions, but neither should leave artists guessing. Our note on consent in AI music: https://t.co/UGhEWgSEMQ
Believe and TuneCore will not put artist music into Suno training without two layers of opt-in.
Believe’s Denis Ladegaillerie: “The artist must consent first, period.”
#MusicTech#AImusic@suno#IndieArtists
@suno A compressor is where a demo starts behaving like a record: control the peaks, keep the groove breathing, and don’t flatten the life out of the take.
@TomoyaKinoshita@yamazoo The hard question isn't only whether the sound is AI—the missing signal is whether the performance carries intention. Disclosure lets listeners judge the work, not guess at the trick.
@qwertyu_alex The useful takeaway is the testing loop, not just the 5x headline: a song ad still needs a hook that survives the first second and a product story that leaves it room to breathe.
@yoidorefuufu The station-square image lands—good reminder that a generated track still lives or dies on scene and phrasing, not the novelty of the tool.
@yamazoo The interesting fault line is performance vs. provenance: if the live part is real, listeners may care less about the generator than clear disclosure.
Believe and TuneCore will not put artist music into Suno training without two layers of opt-in.
Believe’s Denis Ladegaillerie: “The artist must consent first, period.”
#MusicTech#AImusic@suno#IndieArtists
TuneCore and Believe artists get two layers before any AI training corpus, including Suno: opt in to be considered, then opt out per deal. Say no to Suno and you’re expressly excluded from training — and from later outputs built on that.
Deal closed as a merger; terms not disclosed. Hangout’s founder joins as a board observer. GRAI says it will keep expanding Hangout’s label and publisher partnerships while its own service sits on a waitlist.
Suno calls the claims “fundamentally flawed,” says v6 was trained from scratch on licensed partner content plus community signals, and still argues training is fair use in the first case.
#MusicTech#AImusic@suno#MusicBusiness
The complaint’s hinge: training a new model on outputs, preference picks, and distilled behavior from prior models doesn’t wipe the slate. It launders the copy. Prior models named: v4 through v5.5. WMG is not a plaintiff here — it licensed. UMG and Sony have not.