A Chinese group called Supply and Marketing Cooperative Band is heading out on a national club tour while still fighting a summer-long accusation that its music is made by AI, according to a Music Finance report republished by 36Kr on 24 September. A second batch of tickets for the Opening for Business Today tour went on sale on 22 September, adding Wuhan on 24 October and Shijiazhuang on 31 October to an earlier Jinan date, at 128 yuan early-bird and 168 yuan standard.
The group's first festival set came only on 30 August, a 40-minute opening slot at a beach music carnival in Rizhao, less than two months before the tour went on sale. Reports described a relaxed, folksy urban sound built on simple instrumentation and story-like singing, and the band's Weibo account marked the day by saying it had left its first song in the evening breeze on the beach. Yet on the platform where clips of the show circulated, the top related search was for a real band that writes songs with AI, and comment threads split between fans and sceptics who said the live vocals sounded nothing like the records or that the members barely seemed to know each other.
The doubts began in early July, when listeners posted that they felt hurt to learn a band they loved might be AI-synthesised, and others dismissed it as a stitched-together "Frankenstein band" borrowing from many original musicians. The group has never conceded that its songs are AI products.
Public records give a partial picture. The act was set up on 25 June 2024 under the label A Sparrow, with Jiangxin Wuxian listed as the original rights holder. It has about 114,000 followers on NetEase Cloud Music, more than 100,000 on Qishui Music and 26,000 on QQ Music. Its most popular track, Let Time Fade It Away, has been saved over 770,000 times on NetEase and leads its catalogue on Kugou and Shazam; the song credits a writer named Bianyuan and names separate people for arrangement, mixing, guitar and production. Other songs such as I Hear It and Let All Thoughts Go with the Wind also rank highly on several platforms' charts.
What most fuels suspicion is volume. Music Finance counted at least 74 releases on NetEase since mid-2024, mostly two- or three-track singles pairing a song with instrumental and DJ versions, for 133 songs in all, plus more than 12 EPs in 2026 alone on Shazam. Between early July and mid-September a new title arrived every week or two, a pace the report calls unusual for a group presenting itself as human. Titles in that stretch included Mr. Jiang and Miss Wang on 3 July, Lonely Train on 31 July, Ripples on 27 August and Give Me Back Myself on 14 September, spanning urban ballads, folk and pop.
There are plagiarism claims too. Liu Ang, singer of the Aotuti Band, posted videos alleging that the song How How How copies his band's Dust Smoke, setting similar passages side by side. The report describes the wider worry as "splicing": AI tools recombining the melodies, lyric structures and arrangements of many existing songs so that the result matches none of them outright while echoing all of them.
The band has answered twice. It first insisted it was not AI and was preparing to tour, asking whether AI even existed in 2023 and 2024. On 2 September it posted that it did not want to be discussed, judged or defined, and put its output down to a large stock of songs written before the group formed; in another post the same day it asked critics to leave it alone, conceding its musical level was modest.
A senior producer consulted by Music Finance said the vocals on the released tracks sound like a real singer, with natural breathing, diction and shifts in feeling, so calling the group an AI band is inaccurate. In his view it looks more like songs whose words and music were generated with AI tools and then sung by people. He added that high output proves nothing now that AI is common in arranging, mixing and drafting; the real question is how plagiarism is established.
NetEase Cloud Music told reporters it does not let users gain verified musician status with AI works, but acknowledged it cannot reliably flag every AI track in advance, because the technology changes quickly and many songs mix machine output with manual edits.
Part of why the case has drawn so much attention is that the songs are genuinely liked: they are widely saved and became a minor hit on short-video platforms such as Douyin. The report suggests that combination of popularity, a heavy release rate and familiar-sounding melodies made the AI label easy to attach.
The report uses the case to raise questions the industry has not settled: whether platforms or artists carry the blame when unlabelled AI-assisted songs go viral and listeners find out for themselves, whether musicians should have to prove how they worked, and whether someone singing AI-written material on stage is a musician or a performer of a product. It points to listener backlash elsewhere, including anger at AI-made posters sold as souvenirs by the Can Music Festival and criticism of science-fiction writer Hao Jingfang after she said AI was used in her new novel.
It then turns to Suno, whose v6 models launched on 9 September as its first generation trained only on music licensed from partners including Warner Music, BMG, Believe and TuneCore. The range comes as v6, v6-wild and v6-mini, with the first two reserved for Pro and Premier subscribers and v6-wild pitched at producers as a brainstorming aid. Suno Studio can split a song into up to 12 aligned stems, apply six-band EQ, align tempo, strip reverb and export MIDI into programs such as Ableton, Logic Pro and FL Studio; product chief Jack Brody said revenue sharing with the labels began on launch day.
Early reception has been poor. Long-time subscribers on Reddit describe v6 output as muffled, flat and overcompressed, some are cancelling, and the older models cannot come back because they were withdrawn under label pressure. Suno has not explained the change, but the report notes that v6's training set has not been fully disclosed and is evidently smaller and less varied than before, which many in the industry link to output quality. One user who had released two albums made with v5.5 said listeners could barely tell its vocals and orchestration from real ones, yet two days of free credits spent on v6 with the same prompts gave him only dull, generic tracks.
Suno has settled with Warner Music and signed a licence, while its disputes with Universal and Sony go on. The report's conclusion is that every music generator, Suno and Udio included, is at root a product of blending existing songs, so its output inevitably carries traces of what it learned from, the same concern the Chinese band's critics keep raising.
Sources · 1