Photo credit: Alexander Shatov
With With the proliferation of artificial intelligence (and sound-like titles) in the music space and beyond, the RIAA has officially subpoenaed Discord to identify the “AI Hub” server users who allegedly breached protected media.
This latest attempt to stem the tide of alleged AI-powered infringements was only recently brought to light in the actual DMCA subpoena as well as in a report by TorrentFreak, who also spoke to one of the admins of the said server. First and foremost, legal documents show that the RIAA sent a letter to Discord at the end of May demanding the complete deletion of the AI Hub.
“We have learned that Discord operates and/or hosts the below Discord servers on its network,” the industry representative explained on the AI Hub. “This server(s) is/are designed to infringe our members’ copyrighted audio recordings by offering, selling, linking, hosting, streaming and/or distributing files containing our members’ audio recordings without permission.”
And in addition to the above request to “remove and/or disable access to this Discord server,” the RIAA also went further and disclosed the allegedly infringing works (Mariah Carey’s “Always Be My Baby,” Shakira’s “Whenever, Wherever” and Frank Sinatras). “In the Wee Small Hours of the Morning” (including “In the Wee Small Hours of the Morning”) and related links.
While Discord has so far opted against shutting down the server – which currently has more than 145,000 members and includes all manner of posts related to “how-to guides and AI models,” the release said TorrentFreak – The RIAA sent another “server removal request” on June 12 before proceeding with the subpoena detailed above, according to legal documents.
According to the subpoena, Discord has until June 30 to either oppose the request or provide “information sufficient to identify the alleged infringers,” including their names, addresses (both physical and IP), and more.
As for the specific reasons for the shutdowns and subpoenas, the aforementioned AI Hub admin said: spoke with TorrentFreak expressed the belief that “links to full datasets” – not necessarily language models themselves – were (and are) the culprit. The AI Hub rules also prohibit violating records, and the responsible administrator has emphasized this in an announcement to the server members.
“Please do not upload any datasets or copyrighted material to the server,” the person wrote, adding that the AI Hub administrators only learned of the RIAA’s demands from the data subjects and not from Discord itself. “We have zero tolerance for things like this. Thank you very much.”
Of course, this isn’t the first time the RIAA has officially attempted to discover the identity of Discord users for copyright reasons, and of course the continued development of AI, particularly in music, means the organization is cracking down on certain of the applications of the the underlying technology shows little sign of slowing down.