Infringement risks of AIGC in entertainment industry (Part 2)

Following the previous article’s analysis of artificial intelligence-generated content (AIGC-related) infringement risks in the film and music industries, this instalment turns to corporate governance and compliance responses. As AI becomes more deeply embedded in scriptwriting, soundtrack production and digital human development, traditional talent management and licensing agreements are no longer sufficient to address new-use cases.

The resulting issues spanning personality rights, copyright, data provenance and platform compliance are becoming material legal risks for both day-to-day operations and capital markets activity.

For entertainment companies pursuing financing, listings or other transactions, weaknesses in AIGC-related intellectual property and data compliance are now a recurring focus of due diligence by banks and regulators. This article examines practical steps companies can take to strengthen their AIGC compliance frameworks.

Contractual governance

Talent management and performance agreements. Companies should include dedicated AI clauses in their standard contracts, clearly stating whether production teams or third parties may collect and use an artist’s performance materials, including voice, facial expressions and body movements, for AI database building and model training.

Where such use is authorised, the contract should strictly define the permitted scenarios for any digital replica, such as post-production dubbing for the same project or promotion of specified spin-off products, as well as the term, territory and revenue sharing arrangements tied to that authorisation.

Companies should also consider an “artist exit” mechanism, under which the artist may unilaterally revoke consent for use of the AI replica in the event of defined ethical concerns or a major reputational crisis. That helps balance commercialisation of the technology with protection of personality rights.

Commissioned works agreements (screenwriting, scoring and outsourced production). Contracts should require express representations and warranties on AI-assisted creation. External teams should be obliged to accurately disclose whether AIGC tools were used, to what extent, and which portions of the deliverables were AI-generated.

They should also be subject to stringent indemnity obligations for IP defects, with the agreement making clear that if infringement arises from the contractor’s improper use of AIGC tools, all direct and indirect losses will be borne by the contractor.

For music and other works closely tied to personal creative identity, the contract should also specify how authorship or attribution will be reflected where AI-generated content is involved.

Compliance controls

Internal policies. Producers should adopt detailed rules governing the use of AIGC tools and maintain a dynamically updated whitelist of approved systems. In screening tools for use, three criteria deserve particular scrutiny:

1. 1. The legality and transparency of the underlying training data;
   2. Whether the vendor offers copyright infringement indemnity; and
   3. Whether the user terms clearly grant commercial-use rights while undertaking not to use enterprise inputs for further model training.

Robust ex ante ex ante (based on forecast) controls are essential to prevent staff from using open-source models of uncertain provenance, or without adequate copyright safeguards in commercial projects.

Procurement and pre-launch review. Companies should prioritise enterprise-grade commercial AI tools, especially those that offer contractual copyright indemnities. In addition, copyright clearance should be brought forward into the content development stage, with rigorous human review and comparison procedures to identify whether AI-generated outputs are substantially similar to well-known IP or existing protected works.

Virtual persona assets

When multi-channel network (MCN) agencies incubate virtual idols or produce AI-generated short-form video, the IP in a virtual character is often entangled with the personality rights of the performer behind it. Those rights should be separated by contract at the outset of the relationship.

The key is to define clearly that ownership of the virtual persona and rights to its subsequent operation belong to the agency, helping prevent a creator who leaves or terminates the contract from taking the digital asset with them, or later asserting voice-related infringement claims against the company. This is critical if early-stage investment in developing the virtual persona is not to become a sunk cost.

Data sources, and traceability

As the technical starting point of the ecosystem, foundational model providers must establish fully lawful and compliant data acquisition mechanisms. Priority should be given to public domain material, properly licensed commercial databases, or revenue sharing arrangements negotiated with rights holders.

On the product side, platforms should also fulfil their compliance obligations by strengthening content filtering and watermark tracing measures, including digital watermarking, to improve the identifiability and traceability of generated content and reduce the risk of AI being used for deepfakes or infringing outputs.

AI is improving content production efficiency and is becoming a central driver of structural change across the broader entertainment industry. Yet, AIGC has not altered the basic logic of rights allocation; it has primarily multiplied the settings in which rights conflicts can arise, placing new pressure on existing legal rules to adapt to emerging applications.

Entertainment companies need to move more quickly towards digitalised and granular risk control systems. For a business preparing for an IPO or other capital markets activity in particular, the legality of data sourcing and the robustness of IP compliance in AIGC workflows have become issues of direct concern to investment banks and regulators, and may directly impact its capitalisation process.

Companies, therefore, need to adjust their business models early and embed compliance planning in advance. Only by striking a workable balance between technological benefits and legal risk can they build a durable competitive moat.

At the same time, more targeted legislation and regulatory refinement remain necessary. There is also a strong case for continued progress towards cross-border rule co-ordination and registration mechanisms better suited to the copyright protection challenges of the AIGC era.

Jiang Shen is a partner and Xu Wen is an associate at Jingtian & Gongcheng

Read more about Generative AI

---

Infringement risks of generative AI in entertainment industry (Part 1)

Widespread adoption of generative artificial intelligence (AI) in the film, television and music industries in recent years has not only triggered a structural shift in content production but also posed a substantive challenge to the sector’s underlying business logic and existing rights framework

---

For more stories about Generative AI, visit law.asia.