AI-Altered Songs May Create a Business in Proving Who Should Get Paid

A widely shared song can leave its creator struggling to answer a basic question. Who used it, how often, and who should be paid?

On 11 November, Indonesian musician and lawmaker Melly Goeslaw raised that problem during discussions about changes to Indonesia's copyright law. She argued that digital platforms should use audio-fingerprinting technology to identify songs, detect their use and help calculate payments to rights holders. She also complained that her royalties from TikTok and Meta did not reflect a transparent account of how often her works were used.

Goeslaw said that nearly half of her songs had experienced misuse on digital platforms. That is her reported personal experience, not a verified industry-wide rate or an adjudicated finding against any platform. Parliament's account, 11 November, 21:42 WIB / 22:42 SGT.

Bisnis reported that Goeslaw also complained about AI-based alterations of her music. Separately, the parliamentary account records her concern about material reappearing after removal. These are attributed complaints, not an independent audit of every upload. Bisnis, 11 November 2026, indexed report. The challenge is not merely identifying the original song. It is following altered copies, documenting their use and proving which rights apply.

Now consider a separate discussion from the same parliamentary process. Lawmaker Umbu Kabunang Rudi Yanto Hunga raised concerns about suspected misuse of traditional Sumba textile motifs on lower-priced marketplace products. The parliamentary report describes allegations, not a court ruling. He questioned why creators should carry the entire burden of finding suspected infringements. Parliament's textile report, 11 November, 22:10 WIB / 23:10 SGT. Music and traditional textiles are very different businesses. Yet their creators face a related commercial problem: products can spread online much faster than owners can document their use and pursue payment.

This analysis connects five developments: AI makes digital content easier to alter; social-media sharing spreads versions quickly; creators complain about incomplete usage information; policymakers are considering stronger copyright protection; and online marketplaces carry physical as well as digital products alleged to imitate creative or traditional designs. The legal rights must be established separately. We think the next opportunity may belong to companies that prove usage and reconcile payments, not simply companies that detect copied content. Audio identification already exists as a commercial technology. Detecting a matching sound is therefore not a new invention, and a match alone does not establish ownership, infringement or an amount owed. The more interesting business is completing the process: Find the relevant use. Establish who owns the rights. Check whether a licence exists. Record the evidence. Determine whether payment is due. Help the owner resolve missing payments or repeated misuse. A detection alert without a reliable answer may create more work for the creator rather than solve the problem.

We see four possible businesses.

Music-use auditing and royalty reconciliation could help musicians, labels and publishers compare usage evidence against their licensing and payment records. Customers would pay for a measurable result, such as identifying an underreported use or resolving an incorrect payment, rather than for another dashboard.

Reupload monitoring and evidence preparation could serve rights holders whose altered songs repeatedly appear on new accounts. The service could maintain records, prepare authorised complaints and track whether a platform actually resolves them.

Licensing assistance for social-media sellers could help merchants and agencies check that music used in advertising, livestreams and promotional clips is properly licensed before publication. The business case depends on whether merchants value reduced disruption enough to pay for the service.

Design-copy monitoring for traditional craft producers could extend similar evidence methods to photographs and listings. The applicable rights need separate checks: Indonesia distinguishes new original motifs, industrial designs, traditional cultural expressions and geographical indications. Traditional community motifs cannot simply be treated as one seller's privately owned registered design. A similar-looking textile does not automatically prove infringement. DJKI's explanation, 12 January 2026.

There is political pressure for change, but the reviewed evidence does not establish a new law requiring every platform to buy fingerprinting or auditing services. The proposal should not be confused with existing copyright protections. In fact, the resistance is part of the signal.

On 11 November, Reuters reported that Meta, TikTok and Google were challenging other proposed copyright provisions, including payments for news links and rules affecting AI. Those provisions are legally different from audio fingerprinting. Nevertheless, they show that the wider debate over identifying rights, distributing content and assigning responsibility is becoming commercially contentious. Google warned that compulsory compensation for displaying news links could reduce access to journalism. Meta argued that the payment requirement could cost publishers traffic and revenue. TikTok separately sought protection for user-generated-content platforms and a distinction between AI-created and AI-assisted material. These are the companies' positions, not established effects of an enacted law. Reuters, 11 November, published 07:13 EST / 20:13 SGT and updated 07:20 EST / 20:20 SGT.

That disagreement creates an important choice. An entrepreneur could wait for compulsory regulation, hoping platforms will eventually need new compliance systems. Or an entrepreneur could try selling a service directly to creators who can already demonstrate a costly problem. We think the second route offers a more realistic early test.

A small rights-auditing business could approach ten music rights holders and offer a four-week paid investigation using material the customers are authorised to share. The service would compare each customer's catalog, licensing information, available platform records and suspected unauthorised uses. It would document discrepancies without promising that every detection leads to recoverable royalties. The provider would need lawful access to records, clear client authority and qualified legal review where rights or claims are disputed. Monitoring is not permission to scrape restricted data, issue claims or collect royalties without the required authority.

An illustrative first trial might look like this:

• 10 paid pilots at IDR5m each: IDR50m revenue.
• Assumed direct delivery costs: IDR30m.

An additional IDR15 million in initial setup spending would leave an illustrative IDR5 million cash surplus before other overhead, financing and tax, assuming all ten customers actually pay and delivery spending stays within budget. This is a simplified cash calculation, not an accounting-profit forecast.

Entirely hypothetical test economics. No verified customer price, contract, willingness-to-pay study or royalty recovery supports these amounts.

The example is deliberately modest. It shows why attracting paying customers matters more than claiming a market worth millions.

For investors, the named public companies are currently better understood as businesses exposed to possible regulation, not confirmed winners of a new market. The original research recorded Meta Platforms (NASDAQ: META) at US$640.01 at 13:37 EST and Alphabet Class C (NASDAQ: GOOG) at US$291.67 at 14:07 EST on 11 November. PRICE UNVERIFIED: the precise price-and-minute pairs could not be reproduced from the accessible historical records during this review. They are retained only as unverified research observations, not current quotes, closing prices or valuation evidence. No price move can reliably be attributed to these hearings from the evidence reviewed. Meta historical record; Alphabet historical record.

Smaller technology companies may eventually sell rights-management or content-identification services, but no newly funded Indonesian contract or listed company with proven direct revenue from this proposal was identified. The immediate opportunity is therefore more convincing as a small commercial service test than as a reason to buy a particular share.

The decisive question is whether creators will pay today to improve their information, protect licensed works or pursue documented losses. If they will, a viable business could form before legislation is completed. If they will not, public anger and parliamentary attention may never translate into profitable demand.

Report this passage