AI Tool Labels & Strikes Down Original Soundtrack - YES - not an AI-generated one!
From Abhivardhan, our Chairperson
Seems like the AI-enabled labelling and Intellectual Property Law issues have taken an interesting turn in India.
I am sharing a video of Viepsa Arora, which might interest people a bit.
So, days ago Viepsa on her instagram profile had posted about the deletion of some of her soundtracks she had developed as an independent music artist for some time now. She is one of the few and emerging Indie talents from Bharat, who wishes to create their own economic spaces in the largely monopolised music industry as we know it.
The deletion of Viepsa's soundtracks - as she had originally lamented about it in this reel was done because some random AI tool was utilised in detecting AI-generated content.
As Viepsa claims, the content should not have been deleted and retracted like that. There have been issues of fals positives in the world of art (example: https://www.reddit.com/r/aiwars/comments/1aqysrw/ai_art_detectors_and_the_dangers_of_false/?rdt=39180) where original upscaled (or downscaled) artistic works, made by hand (not AI) of real artists have been wrongfully or misleadingly detected by these AI tools as “AI-generated”.
This reminds me of the International Scientific Report on AI Safety published as a part of the Seoul AI Summit of May 2024 (read my analysis of the report: https://www.indicpacific.com/post/ai-seoul-summit-2024-decoding-the-international-scientific-report-on-ai-safety) where it was specified that AI techniques built to label & track deepfakes and other synthetic content are not that effective.
Well, I'd just say that as I had recommended in India's first privately proposed AI bill, https://aiact.in, we should make AI detection practices of forensic value, both human-based and technical, open source. In addition, we should do some versioning of these detection practices in an open-source repository, so that anyone, be it a lawyer, a bureaucrat, a doctor, an artist, a citizen, a consumer or anyone could track, and make informed decisions.
What happened with Viepsa shows why the IT Rules of 2021, are needed, and why would a Digital India Act in future would be needed if not an AI act. We can't wait for some crazy US Senate/House hearing to put these companies building these substandard & unaccountable AI tools on the stand.
What do you think?