Artificial intelligence and intellectual property rights: the situation in Thailand
Suebsiri Taweepon
Tilleke & Gibbins, Bangkok
Artificial intelligence (AI), particularly generative AI, is transforming how businesses innovate, create content and commercialise intellectual assets. While AI offers significant opportunities, it challenges legal concepts developed around human creators and inventors, raising complex questions regarding ownership, infringement and protection.
AI can be both a creator and consumer of intellectual property (IP). AI systems are trained using vast quantities of potentially copyrighted data, while their outputs may resemble existing works or generate content that would traditionally qualify for IP protection if created by a human. Simultaneously, AI technologies themselves represent valuable commercial assets that businesses seek to protect. These overlapping roles have exposed gaps in legal frameworks worldwide.
Thailand’s current IP framework predates modern AI and provides limited guidance on AI-generated works, AI training practices and AI-generated inventions. This article reviews the existing legal framework, explores key IP issues arising from AI and discusses practical considerations and future developments.
The current legal framework
Thailand currently has no AI-related IP laws. Existing legal principles generally apply to AI-related activities, but there is limited guidance on AI model training, no text and data-mining exceptions and little judicial precedent that could be applied to AI- and IP-related legal issues.
This position is broadly consistent with the wider Southeast Asian landscape, where copyright protection remains closely linked to human authorship and legislation is largely silent on AI-generated content. As a result, AI developers, organisations deploying third-party AI tools, content creators using generative AI and investors assessing AI-related legal risks must rely on contractual relationships to ensure legitimate use and mitigate legal exposure. The absence of clear AI-specific rules elevates the importance of contracts and internal governance, making contractual arrangements the primary mechanism for allocating ownership, liability and risk. Businesses should consider legal protections early on in the AI adoption process rather than waiting for regulatory certainty.
Regarding IP disputes, Thailand’s Civil Court issued a regulation, effective since 17 November 2025, governing the use of AI in the preparation of pleadings and other documents submitted to the court. The Civil Court’s rule is consistent with guidance previously issued by the President of the Supreme Court in September 2025, which was intended to promote accuracy, transparency and public confidence in the administration of justice. The Civil Court’s rules require parties and their counsel to disclose any AI-generated content submitted to the Civil Court and to verify and certify its accuracy. IP cases in Thailand must be filed at a specialised court called the Central IP and International Trade Court (IP&IT Court). It is likely that the IP&IT Court will issue similar rules related to the use of AI, but as of the date of this article, no such announcement has been made.
Copyright considerations
Among all of the available IP protections, copyright presents the most significant challenges for AI developers, users and content creators.
AI-generated works and human authorship
Copyright law in Thailand, as in many jurisdictions, is founded on human authorship. While the law protects original literary, artistic, musical, audiovisual and software works, it does not expressly address works generated autonomously by AI systems.
International developments point in a similar direction. The US Copyright Office has confirmed that copyright protection generally requires human authorship, while acknowledging that AI-assisted works may qualify where sufficient human creative contribution exists. Although not binding in Thailand, this provides a useful indication that it is very likely that Thai regulators and Thai courts may adopt a similar approach.
For businesses and creators, human involvement in the creative process should be carefully documented. Records demonstrating how individuals selected, arranged, modified or contributed to AI-generated content may become important for establishing ownership or enforcing copyright claims.
Ownership of AI-generated content
Even where sufficient human involvement supports a copyright claim, ownership may not be straightforward. Questions may arise as to whether rights belong to the individual user, the organisation directing the project, the employer or another party. Contractual arrangements can resolve many of these issues, but businesses should review ownership provisions carefully before relying on AI-generated content commercially. For example, under Section 9 of the Thai Copyright Act B.E. 2537 (1994), the default rule is that an employee owns the copyright in a work created in the course of employment, unless the parties have agreed otherwise in writing. Therefore, businesses must ensure that employment agreements adequately address the ownership of copyright works created using AI tools.
Training AI models using copyrighted content
A more pressing issue concerns using copyrighted material to train AI models. Modern AI systems depend on vast datasets containing text, images, code, audio, video and other content, much of which may be protected by copyright.
Unlike Singapore and Japan, Thailand does not provide a broad text- and data-mining exception permitting the use of copyrighted works for AI training purposes. Reproducing copyrighted works to train AI models may constitute infringement unless authorisation has been obtained or a specific statutory exception applies.
This creates significant challenges. Rights clearance at scale is time-consuming, expensive and sometimes impossible. The lack of clear legislative guidance creates uncertainty, exposing developers to potential infringement claims, injunctions and damages.
Developers may also face allegations that model outputs reproduce protected content or constitute unauthorised derivative works. Although these questions remain largely unresolved under Thai law, they are increasingly subject to litigation elsewhere and such approaches are likely to influence future discussions in Thailand.
Data scraping presents additional issues. Publicly accessible information cannot necessarily be copied into training datasets without legal consequences. Businesses should not assume publicly available content is automatically available for AI training, as most such content is protected under copyright law in most jurisdictions and requires explicit consent for use, particularly for commercial purposes.
Managing copyright risk
Given these risks, businesses should undertake careful due diligence concerning training data origins, terms governing third-party AI tools and the licensing status of materials used in model development. Organisations should review whether AI vendors use customer inputs for model training, which rights users retain in regard to any outputs and whether indemnities are available for infringement claims.
Companies commissioning customised AI solutions should seek contractual protection, including warranties regarding training data sources and indemnities for IP infringement. Robust contractual drafting remains one of the most important risk-management tools available.
Patent protection for AI innovation
Patents remain important for protecting AI innovation. Novel processes, technical applications, system architectures and implementation methods may be eligible for patent protection if they satisfy the relevant novelty, inventive step and industrial applicability requirements.
However, AI-related applications often encounter challenges because mathematical methods and abstract ideas are generally not patentable. Applicants are more likely to obtain protection where an AI innovation connects to a specific technical application, such as medical diagnosis, industrial automation, logistics management or fraud detection, rather than the underlying algorithm alone.
Businesses should carefully weigh the benefits of patent protection against the requirement for public disclosure. In fast-moving AI markets, where technologies may be difficult to reverse engineer, some organisations may find trade-secret protection more effective. As a result, AI businesses often evaluate patent and trade-secret strategies in parallel.
Another unresolved question concerns inventorship for AI-generated inventions. Thailand’s patent system assumes inventors are human. The Patent Act B.E. 2522 (1979) provides no clear mechanism for recognising an AI system as an inventor, leaving uncertainty in circumstances where AI plays a substantial role in the inventive process.
Trademarks and AI branding
As competition within the AI sector intensifies, trademark protection is important for establishing brand identity. AI developers should consider trademark protection for platform names, software products, AI-powered services and associated logos.
One challenge is distinctiveness. Terms such as ‘AI’, ‘smart’ and ‘tech’ are widely used and may be regarded as descriptive or generic. Under Thai trademark practice, such terms are unlikely to be registrable unless combined with sufficiently distinctive elements. Businesses should prioritise original branding strategies that clearly differentiate their products from those of their competitors.
Trademark issues may also arise as a result of generative AI itself. AI tools can create logos, brand concepts and marketing materials quickly, but businesses should not assume AI-generated branding is automatically available for unrestricted use. Clearance searches remain essential to ensure newly created names and logos do not infringe existing rights.
Choosing effective trademark classification in a trademark application is also highly relevant and critical for trademark protection. For an AI software as a service (SaaS) business, Class 42 is generally relevant to non-downloadable software platforms, while Class 9 may apply to downloadable software, apps, software development kits (SDKs), application programming interface (API) clients or AI-enabled devices. Class 35 may also be relevant for business, marketing or commercial data services. Where AI offerings are sector specific, such as AI used in Fintech, healthcare, logistics, legaltech, robotics or autonomous vehicles, the relevant industry class should also be considered.
Trade secrets: protecting confidential AI assets
For many AI businesses, trade secret protection is as important as, or in some cases more important than, patents. Unlike patents, which require disclosure and have a finite duration, trade secrets can potentially protect valuable information indefinitely, provided confidentiality is maintained. Valuable AI assets protected as trade secrets may include source code, proprietary datasets, algorithms, model weights, training methodologies and other confidential know-how.
Unlike patents, trade secrets require no registration and are protected through confidentiality measures, such as access controls, security policies and confidentiality agreements. Thailand’s Trade Secrets Act B.E. 2545 (2002) provides the statutory framework for trade secret protection. As generative AI becomes more widely used, businesses should adopt clear AI-use policies to prevent unauthorised disclosure of confidential information through AI tools.
Data protection and privacy considerations
AI datasets may contain personal data, and such use should be assessed not only from an IP perspective, but also under Thailand’s Personal Data Protection Act B.E. 2562 (2019) (PDPA). The PDPA, which has been fully in force since 1 June 2022, regulates the collection, use and disclosure of personal data and recognises data subjects’ control over how their personal data is collected, stored, processed and disseminated. Accordingly, a licence to use a dataset for copyright purposes does not, by itself, resolve PDPA compliance where the dataset includes identifiable individuals.
Organisations should assess whether training, testing or analytics data contains personal or sensitive personal data, identify the appropriate lawful basis and provide clear privacy notices. They should also implement suitable governance, security, retention and data-transfer safeguards, particularly for cross-border transfers.
Future developments
Thai policymakers have begun exploring how the legal framework should evolve in response to AI. Discussions include possible AI-specific regulation, copyright exceptions for text and data mining and broader modernisation of existing laws. Thailand's Department of IP and Electronic Transactions Development Agency (ETDA) may issue guidance or propose legislative amendments as AI adoption accelerates, although the timing of any such reforms remains uncertain.
Thailand’s approach is likely to be influenced by developments in the US, European Union, Singapore, Japan and China, as well as regional harmonisation efforts within ASEAN. As international approaches diverge on AI training, copyright exceptions, transparency requirements and liability standards, businesses operating across jurisdictions face greater compliance complexity. Monitoring these developments remains essential for organisations with regional or global operations.
Notably, on 2 July 2026, the ETDA released a revised draft act on AI for public consultation. The draft adopts a risk-based regulatory framework and would apply to AI activities affecting persons in Thailand, including certain activities conducted outside the country. Key provisions address the risk classification of AI systems, governance requirements for high-risk AI, transparency obligations for AI-generated content, regulatory oversight mechanisms, liability for AI-related harm and enforcement measures. The draft also includes provisions on regulatory sandbox programmes, self-regulation and phased implementation.
Conclusion
AI creates unprecedented opportunities for innovation, while exposing the limitations of IP frameworks designed for human creators. In Thailand, unresolved questions surrounding authorship, AI training, inventorship and ownership continue to generate legal uncertainty. Until the law develops further, businesses should focus on rights clearance, contractual safeguards, documentation, internal governance and proactive risk management. Organisations combining technological innovation with thoughtful IP strategies will be best positioned to capitalise on AI, while protecting their intellectual assets.