How will the most important labels overcome the copyright risk from AI music? By turning to probably the most highly effective weapon out there: AI itself.
Enable us to make a prediction.
In 5 years, the most important music firms won’t solely be scouring the net for AI infringement, they will even be issuing authorized letters on to the perpetrators… utilizing AI.
Sound far-fetched? It isn’t. It’s sitting inside a pair of patent functions printed by the US Patent and Trademark Workplace on February 12, 2026.
MBW unearthed the filings whereas researching our latest story on the patent portfolio being constructed by Music IP Holdings, the entity shaped final yr by way of Common Music Group‘s partnership with IP asset administration agency Liquidax Capital.
That earlier story reviewed three filings within the portfolio masking multi-stage approval and managed distribution of AI-generated by-product works. Nashville-headquartered MIH has mentioned it holds “greater than 60 protected improvements with quite a few extra expertise households and portfolios underneath growth.”
The 2 filings we’re taking a look at right here seem like a part of that broader portfolio.
Titled “Media Rights Platform Techniques and Strategies” and bearing publication numbers US 20260044581 A1 and 20260044583 A1, they record Music IP Holdings as sole applicant and Daniel Drolet, Liquidax’s founder and MIH’s Chairman and CEO, as sole inventor. That may be a completely different sample from the by-product works filings coated in our earlier piece, which embody UMG staffers Chris Horton, Jeremy Uzan, and Sion Elliott among the many co-inventors.
Each filings have been lodged with the USPTO on October 16, 2025. As with the sooner derivative-works filings, neither MIH nor UMG has publicly disclosed whether or not or once they intend to deploy or license the expertise described, or to whom.
What the filings describe is a sprawling end-to-end “media rights platform” that sits between rightsholders, generative AI programs, and the tip customers who wish to immediate AI to create by-product works from copyrighted music. Nevertheless it’s the enforcement equipment bolted onto the platform that ought to make AI companies take discover.
4 parts of the filings stand out, plus a fifth that borrows from a playbook the stay live performance ticketing enterprise already is aware of nicely.
1. A licensing chatbot that plugs immediately into ChatGPT, Claude and others
On the coronary heart of the filings is a system known as the “Copyright Licensing Chatbot” (referred to within the patents as element 190), described as an “agent” or “plugin” that plugs immediately into present massive language mannequin platforms – the filings explicitly name-check ChatGPT and Bard, but in addition reference Anthropic’s Claude 2, Google‘s LaMDA and PaLM, Hugging Face’s BLOOM, Nvidia’s NeMo, XLNet, Cohere and others.
The bot is designed to interrogate potential licensees about industrial vs. non-commercial use, timeframe, and geographical scope earlier than both clearing or escalating the request.
The filings state that the chatbot “questions the potential licensee on various utilization attributes to determine each the suitable licensing fashions and to verify that the proposed utilization of the content material is aligned with the artist’s rules and general necessities that have to be met earlier than their materials shall be licensed and utilized in a by-product work.”
Danger profiling is dealt with by machine studying fashions skilled on historic licensing information and infringement patterns. Low-risk, non-commercial queries will be cleared in-conversation. Excessive-risk requests get escalated to people.
2. An AI crawler that hunts infringement – and sends its personal cease-and-desists
That is the piece which may matter most for MBW readers.
The filings describe what they name an “LLM Agent Copyright Crawler” – a system that “works throughout the Web and interacts with Copyrighted Spinoff Content material created by Customers of LLM’s”.
The filings describe the crawler as sampling content material streams from the open internet, detecting digital watermarks embedded in photographs and audio utilizing machine studying, and cross-referencing that materials towards IP licenses “at the moment in drive”.
The filings are specific about what the agent does when it finds a mis-match.
The filings state that the “LLM Agent” can join immediately with a supply of supplies and “suggest licensing phrases”; it will possibly thank customers for having “correctly licensed utilization”; and – critically – it will possibly ship “a number of stop and desist letters to consumer or streamer”. It may additionally flag the utilization for “human authorized intervention”.
That’s, functionally, a blueprint for agentic copyright enforcement. A bot that, as described within the filings, would discover infringement, assess threat, and difficulty authorized correspondence, with human legal professionals concerned solely as an escalation tier.
3. An AI mannequin of the rightsholder that predicts whether or not they’d say sure
The platform additionally contains an AI Modeling System (element 180) that fashions the copyright proprietor themselves, utilizing machine studying skilled on “historic licensing choices, authorized precedents, and copyright holder habits patterns”.
The aim: to foretell, earlier than a by-product work is ever created, whether or not the rightsholder would approve it.
The filings describe the system as “approving creation of the by-product work in response to the modeling predicting that the proprietor would approve the request.” Spinoff work creation is then accepted or declined on the idea of that prediction.
4. Digital watermarking and fingerprinting to hint each AI-generated monitor
Underpinning the entire system as described is a Digital Watermarking System (element 170) utilizing unfold spectrum, quantization-based, rework area, and perceptual watermarking strategies.
Paired with it’s a “digital fingerprinting engine” designed to detect the signatures of particular generative AI fashions – what the filings name “GAI-MC” fingerprints, described as “distinctive byte sequences generated by particular AI mannequin architectures”.
The filings describe the system as figuring out “which LLM and music aligned software program created the copyright derivatives as a way to determine what number of derivatives have been created, how a lot cash is owed, how a lot worth was used to coach the LLMs as a way to create the derivatives.”
5. A dynamic pricing engine for AI music licenses
Maybe most intriguingly, the filings describe a dynamic pricing engine that adjusts licensing prices in actual time “based mostly on demand, seasonality and market situations” utilizing reinforcement studying strategies to repeatedly optimize pricing based mostly on “conversion charges, buyer lifetime worth, and market penetration targets”.
The system is designed to mannequin worth elasticity of demand, analyze competitor pricing, and outline “dynamic pricing guidelines and algorithms that regulate costs in real-time based mostly on demand alerts, stock ranges, competitor pricing, and buyer segmentation”.
The filings even ponder the system “dynamically adjusting approval standards based mostly on market components, demand patterns, and seasonality,” which means not simply the worth of a license might flex in actual time, however whether or not a license is granted in any respect.
The filings recommend this is able to be paired with customized subscription plans, bundle choices, loyalty rewards, and tiered pricing for AI music licensees – a mannequin that, to MBW’s eye, at the very least, seems loads much less like conventional music licensing and much more just like the demand-based dynamic pricing already acquainted from the stay live performance ticketing enterprise, utilized to copyright permissions.
The paperwork aren’t merchandise. They’re patent functions. They might by no means ship within the type described, and patent claims typically stretch wider than what will get constructed. As with the remainder of MIH’s portfolio, UMG and MIH haven’t publicly disclosed their industrial plans for the expertise.
However the filings inform us one thing about the place UMG and its patent-licensing associate imagine the business is heading, and the extent of automation they’re getting ready to say IP rights over.
For AI companies already preventing lawsuits from main labels, the route of journey is obvious sufficient. At this time, the cease-and-desist letter is drafted by a lawyer at Latham & Watkins or King & Spalding. Tomorrow, on this blueprint, it might be drafted and despatched by a bot.
5 years, we mentioned. We’ll test again.Music Enterprise Worldwide
