Tech Focus: AI & Production Music —Many Benefits for Broadcast Sports, but Uncertainty Continues

Legal/copyright issues are major obstacle to AI’s widespread on-air and in-venue use

Artificial intelligence is a divisive technology. For broadcast sports, particularly for music used on-air and in venues, it is both embraced and eyed suspiciously.

For instance, former Turner Sports VP Tom Sahara, SVP, production technology, Quintar, a spatial-experience developer, has cited AI’s positives, including assisting audio teams by monitoring multiple audio parameters, such as signal levels and routing. However, he cautions about the technology’s propensity for generating digital “hallucinations.”

Similarly, Karl Malone, senior director, audio engineering, NBC Sports and Olympics, is optimistic about AI’s potential in automating routine tasks, such as some types of submixing, but with a caveat: “as long as there is someone dedicated to the audio design of the production leading it and not using it as a ‘set and forget,’ because the ‘forgetting’ part is where we can run into problems.”

Music, which plays a huge role in all sports, both on the air and on the field, is where much of the controversy is manifest — from creation to playback. And nowhere is it felt more than at the various companies that generate and/or license thousands of music tracks every year to teams, leagues, broadcasters, and streamers, providing the underscores and themes not only for games but for entire productions.

The Human Touch

Production-music companies — many of the largest count sports entities as major clients — are aware of the legal and social sensitivities around the topic.

“As far as we can tell,” says Ron Mendelsohn, CEO/executive producer, Megatrax, “our clients still prefer music produced by humans rather than music generated by machines, especially when the legal rights and copyrights to that music are so questionable. Several clients have asked us to verify that our music is not AI-generated.” It’s “highly problematic,” he adds, that AI-generated music often bases training on copyrighted content.

Another issue is that AI-generated music itself is not entitled to copyright protection. In the U.S., works created entirely by artificial intelligence are not eligible for copyright protection because the U.S. Copyright Act requires human authorship. (Copyright duration is based on the author’s lifespan, which courts and the Copyright Office have decided implies a requirement for human authorship.) Purely AI-generated music, therefore, can be refused copyright registration by the Copyright Office. On the other hand, AI-assisted music, for which a human provides significant creative input and control over the output, may be eligible for protection, at least on the copyrightable human-authored portions.

The Problem

Copyright owners have accused the AI sector of using their vast catalogs as uncompensated training grounds for AI music. According to international music-publishing trade association ICMP, some of the world’s biggest technology companies — Google, Microsoft, Meta, OpenAI, X — have scraped copyright-protected music to train generative-AI systems. According to “extensive” evidence the association provided to Billboard, that has included songs by the Beatles, Mariah Carey, The Weeknd, Beyoncé, Ed Sheeran, and Bob Dylan.

ICMP describes such unlicensed use of digital music as taking place on a “global and highly extensive scale.” It further asserts that the scope of the training is larger than previously acknowledged.

The problem is widespread across all media. It’s also costly: artificial-intelligence company Anthropic has reportedly agreed to pay at least $1.5 billion to settle a copyright-infringement lawsuit over its use of pirated books to train large-language models (LLMs).

APM’s Matthew Gutknecht: “We will use [AI] to help people save time finding our music, but we will not be using it to create derivative works from existing copyrights.”

(Both Sony Music and Universal Music Group, along with Warner Records, are suing AI-music generators Suno and Udio for copyright infringement, alleging that they used vast amounts of copyrighted music without authorization to train their AI models, which can then create similar works. Many in the industry would like to see a workable AI-licensing model emerge, but, as a recent article in Billboard points out, “What would that even look like?” given the legal complexity of the issue.)

“We will use [AI] for audio identification to help people save time finding our music,” Gutknecht emphasizes, “but we will not be using it to create any kind of derivative works from existing copyrights.”

The Bright Side

Stephen Arnold Music’s Whitney Arnold: “It can be a positive creative tool. We’re firmly in the ‘AI as a tool’ camp at the moment.”

Others, however, see potential benefits for AI-generated production music. Whitney Arnold, president, music services, Stephen Arnold Music, says it is “definitely on our radar. We’ve found practical applications for AI-generated music as an ideation tool.” He notes that a client recently sent the company AI-generated vocal treatments to aurally illustrate various moods and styles that the client might wish to pursue.

He concedes that, although AI-generated content can hit the right musical notes, it consistently falls short on emotional accuracy. However, he counters, “in a lot of ways, it can be a positive creative tool. We’re firmly in the ‘AI as a tool’ camp at the moment.”

Others are similarly positive about the technology’s possible impact on commercial-music creation. “Generally speaking, the emergence of AI has had a positive influence on our business,” says Peter Alexander, sales manager, Sound Ideas, citing licensing of the company’s content for LLM training with the contractual caveat that it cannot be used to create competitive content.

That licensing itself, though, is a new and gray legal area. “Each agreement is slightly different, depending on licensee needs,” he notes. He declines to be specific but adds, “We do have ways to contractually protect our proprietary content.”

Arnold is similarly cautious, noting concerns around creation of derivative works from his company’s intellectual property. “[We’re] trying to navigate that side of things so that we’re not preventing creatives from using tools that make their work better and lives easier,” he says, “but, at the same time, [we want to] protect us so that we keep doing this for another 30 years.”

Legal Uncertainty Makes for a Hazy Future

Infringement continues to be a leading concern with AI-generated music, including — and perhaps especially — for broadcast and in-venue sports production. There has already been much litigation around the use of music for sports productions on-air, over streaming, in venues, and on social media.

“If I was a sports team, I would be very concerned about the ownership of AI-created content,” says Arnold, noting teams’ and leagues’ recent pushback on pricing. “It’s the Wild West right now and seems silly to cut corners to try to save money this way — especially when teams are so profitable — [given] the potential to open themselves up to some kind of copyright- or AI-related intellectual-property issue.”

Navigation of the legal aspects of AI-music content will likely remain fraught for some time. Litigation around the use of copyrighted material for LLM training began to be filed barely two years ago, and cases involving music copyrights can last decades.

Morgan McKnight, executive director, Production Music Association (PMA), whose term ends in December, conveys the elusiveness of certainty around the issue: “It’s almost impossible to fully understand the scale or impact of AI-generated music while lawsuits are still being litigated, policies are still being debated, and laws are not finalized. It’s something we’re watching closely and a process we are involved in as it pertains to the protection of intellectual property. The PMA believes in and will continue to advocate for the protection of copyright and human creators.”

As with many other applications, AI’s presence in production music, in some form or another, seems inevitable. However, like other new technologies — and human interns — it may have to practice a bit at the entry level.

“In our opinion, AI-generated music might be adopted by some consumers for UGC videos,” suggests Metgatrax’s Mendelsohn. “However, the professional market — broadcast, streaming, advertising, and so on — cannot risk licensing uncopyrightable music whose origins are legally murky and ambiguous.”