As Universal clamps down on AI music copycats, China proposes new rules to restrict artificial intelligence development

Credit: Shutterstock/Gil Corzo
China's government – led by President, Xi Jinping (pictured) – has proposed that tech firms will need to submit security assessments if they wish to develop AI projects in the country

One thing that most people across the music business – and the political spectrum – seem to be able to agree on right now is that the rapid development of AI is going to be highly transformative for our professional and personal lives.

Recently, over 1,000 signatories, including hundreds of tech, science, and academic leaders, signed an open letter calling on all AI labs around the world “to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4”.

Over in China, the government has drawn up proposals to take the restriction of the development of generative Artificial Intelligence in the country into its own hands.

The proposed measures, published on Tuesday (April 11) by the Cyberspace Administration of China (CAC) and titled Administrative Measures for Generative Artificial Intelligence Services, would apply to the research, development, and use of generative artificial intelligence products within China.

CAC says that tech firms will need to submit security assessments to the so-called “national network information department” before their AI services can be used by consumers.

The measures include AI-powered tech used to generate text, pictures, sounds, videos, codes, and other content “based on algorithms, models, and rules”.

Under the rules, AI service providers in China will be required to ensure that users “scientifically understand and rationally use” the content generated by AI and “not to use the generated content to damage the image, reputation and other legitimate rights and interests of others, and not to engage in commercial hype or improper marketing”.

Plus, the content generated by AI, according to CAC, “should reflect the core values ​​of socialism, and must not contain subversion of state power, overthrow of the socialist system, incitement to split the country, undermine national unity, promote terrorism, extremism, and promote ethnic hatred and ethnic discrimination, violence, obscene and pornographic information, false information, and content that may disrupt economic and social order”.

Adds CAC: “If users find that the generated content does not meet the requirements of these measures, they have the right to report to the cyberspace administration or relevant competent authorities”.

The arrival of the draft proposals to bring tighter controls around the development and public use of generative AI tools in China arrives as the market’s tech giants demonstrate their own AI models to rival the likes of Open AI’s Chat GPT.

On Tuesday, Alibaba Group unveiled its AI language model called Tongyi Qianwen which, in English, means “truth from a thousand questions”.

Baidu launched its own, reportedly underwhelming version of Chat GPT, called ERNIE, last month.

Meanwhile, in the music industry, generative AI and its potential to infringe on copyrighted content is becoming an increasingly pressing concern for major rightsholders.

Universal Music Group, the world’s largest recorded music rightsholder, has reportedly told Spotify and other music streaming services like Apple Music to limit AI-powered tech from ‘scraping melodies and lyrics’ from its artists’ songs.

AI services are trained on existing copyrighted music in order to generate new content, and this is becoming a subject of increasing concern for rightsholders such as UMG, according to the Financial Times.

The FT reports having seen emails sent to platforms by UMG last month, in which it wrote: “We will not hesitate to take steps to protect our rights and those of our artists.”

Additionally, the music company reportedly told streaming services in March: “We have become aware that certain AI systems might have been trained on copyrighted content without obtaining the required consents from, or paying compensation to, the rightsholders who own or produce the content”.

Coincidentally, Universal is part-owned by a China-based tech giant itself, after a Tencent-led consortium took ownership of 20% of UMG in 2020 and 2021, prior to the music firm’s listing on the Amsterdam stock exchange.

The Tencent-led group closed the acquisition of its first 10% in UMG in Q1 2020. An additional transaction, which took place during the first half of 2021, took the Tencent-led consortium’s total holding in UMG to 20%.

The consortium is led by Tencent, but also included the participation of the firm’s majority-owned subsidiary, Tencent Music Entertainment (TME).

TME parent, Tencent Holdings is developing its own AI tools, including natural language processing and speech recognition covering speech enhancement, acoustic/language modeling, and text-to-speech.

Tencent Music, meanwhile, owns “patented voice synthesis technology”, called the Lingyin Engine, which TME, claims can “quickly and vividly replicate singers’ voices to produce original songs of any style and language”.

As of the end of September 2022, TME had created and released over 1,000 songs with human-style vocals manufactured by the Lingyin Engine.

China’s move to control AI development and Universal’s reported push to limit AI from being trained on copyrighted material both arrive amid the rising popularity amongst consumers to use increasingly accessible AI tools to generate music in the style of popular artists.

A viral example of this can be seen in this video posted by ‘entrepreneur and designer’ Roberto Nickson, in which he used an AI audio model of Kanye West (aka: Ye) to turn his own voice into that of the controversial superstar. (Nickson’s very clever, yet rather unsettling video, is a must-watch – check it out here.)

This growing trend was examined by MBW last week, following reports about a YouTuber receiving a copyright strike from the platform for an AI-generated Eminem ‘cat rap track’, which features AI-generated vocals that mimic the star’s voice.

Universal Music Publishing sent the takedown request to YouTube because, according to a source, it believed that the backing music used in the AI track infringed on the copyright of Eminem’s hit Not Afraid.

We pointed out that it wasn’t clear if the supposedly infringing music accompanying the AI-generated vocals was also generated by AI, but noted that if it was, it highlights the problematic nature of AI models being trained on vast volumes of copyrighted music.

Following the reported communications between Universal and music services, the question many in the music business will be pondering today is, could this be the start of a broader clampdown by major music companies on AI-generated music in the style of popular artists?

The answer to that question, we’d suggest, is that it’s likely – especially in the context of MBW’s report on Tuesday (April 11), that analysts are starting to believe that AI may be an existential threat for the major labels.Music Business Worldwide

Related Posts