1. Introduction

AI Patent Analysis Governance in Thailand concerns the legal regulation of artificial intelligence systems used to discover, analyse, evaluate, develop, register and commercialise inventions.

Artificial intelligence is increasingly used by businesses, universities, research institutions and technology companies to:

Identify potentially patentable inventions.

Analyse existing patent databases.

Predict patentability and inventive steps.

Generate technical designs and engineering solutions.

Detect patent infringement.

Identify competing patent applications.

Evaluate patent ownership.

Automate patent examination and classification.

Assess patent portfolios for commercial investment.

Support patent licensing and technology transfer.

These developments raise important questions under Thai civil law and intellectual property law.

For example:

Can an AI system legally qualify as an inventor?

Who owns an invention generated with substantial AI assistance?

Can AI-generated technical inventions receive patent protection?

Who is responsible when AI patent analysis produces incorrect results?

Can a company claim ownership of inventions generated by its employees using AI?

Can an AI system unlawfully reproduce or analyse another company's patented technology?

Central legal principle: AI can be an instrument of invention and patent analysis, but Thai patent law does not currently provide a clear mechanism for recognising AI itself as a legal inventor.

Thailand's principal legislation remains the Patent Act B.E. 2522 (1979), as amended, supplemented by the Civil and Commercial Code, Copyright Act, Trade Secrets Act and relevant contractual principles.

2. Legal Framework Governing AI Patent Analysis in Thailand

A. Patent Act B.E. 2522 (1979)

The Patent Act is the principal legislation governing invention patents, petty patents and design patents.

Three requirements are particularly important.

Section 5 – Patentability

An invention must satisfy three conditions:

Novelty.

Inventive step.

Industrial applicability.

These requirements apply equally when AI is used to develop or analyse an invention.

Section 6 – Novelty

An invention must not form part of the existing state of the art.

Prior disclosures may include:

Published patent applications.

Scientific publications.

Technical documents.

Public demonstrations.

Previously disclosed inventions.

Earlier patented inventions.

AI patent-search systems can help identify prior art, but their results must be independently verified.

Section 7 – Inventive Step

An invention must not be obvious to a person ordinarily skilled in the relevant technical field.

AI-generated inventions therefore require an assessment of whether the claimed technical contribution satisfies the statutory inventive-step requirement.

Section 8 – Industrial Applicability

An invention must be capable of being made or used in an industrial activity.

An AI-generated theoretical proposal without a sufficiently practical industrial application may not satisfy this requirement.

Section 9 – Excluded Subject Matter

Section 9 excludes certain categories, including scientific or mathematical theories and computer programs.

This is particularly important for AI inventions.

A computer program as such is excluded, but an invention involving software may potentially qualify when it produces a patentable technical contribution and satisfies the other statutory requirements.

These statutory requirements are confirmed in Thailand's official patent guidance.

LawPlayer+2

B. Sections 10–12: Inventorship and Ownership

These provisions are especially important for AI-generated inventions.

Section 10 – Inventor's Rights

The inventor has the right to apply for a patent and to be identified as the inventor.

The right to apply may be assigned or transferred by succession, subject to the statutory requirements.

This creates a significant difficulty for fully autonomous AI inventions: an AI system cannot straightforwardly exercise the statutory rights of a human inventor.

Section 11 – Employee Inventions

Where an invention is made in the performance of an employment contract or a contract for specified work, the right to apply generally belongs to the employer or commissioning party, unless otherwise agreed.

This provision is important when employees use AI tools to generate technical inventions during their employment.

Section 12 – Employee Remuneration

The employee inventor may be entitled to special remuneration where the employer obtains benefits from the invention, subject to the statutory conditions.

Therefore, AI-assisted inventions can create disputes about the respective contributions of employees, employers and technology developers.

Intellectual Property Association of Thailand+1

C. Sections 17–28: Patent Application and Examination

These provisions regulate the patent application process.

They are relevant to AI-assisted patent applications involving:

Technical descriptions generated using AI.

Automated patent classification.

AI-generated technical drawings.

Prior-art searches.

Inventive-step analysis.

Disclosure of technical characteristics.

An applicant must ensure that the patent specification accurately describes the invention and satisfies the statutory requirements.

An AI-generated patent description containing fabricated technical information or undisclosed limitations can create serious problems during examination or subsequent litigation.

D. Sections 35, 36 and 36 bis: Patent Rights and Infringement

Section 35 establishes the general 20-year term of an invention patent.

Section 36 grants the patent owner exclusive rights over the patented product or process, subject to statutory exceptions.

Section 36 bis determines the scope of patent protection by reference to the claims, description and drawings.

These provisions are important when AI systems analyse whether a competing product or process infringes a patent.

For example, an AI system may conclude that a competing industrial machine does not infringe because its technical structure differs slightly from a patented machine.

However, the legal determination must consider the actual patent claims and the statutory scope of protection.

(Section 88 amended)+1

E. Trade Secrets Act B.E. 2545 (2002)

AI patent-analysis systems often process confidential technical information.

Examples include:

Unpublished patent applications.

Proprietary engineering designs.

Manufacturing processes.

Research databases.

Confidential AI training datasets.

Commercial research findings.

The Trade Secrets Act is relevant when AI tools disclose, misuse or improperly acquire confidential technical information.

A business should establish appropriate access restrictions, confidentiality agreements and technical safeguards.

F. Copyright Act B.E. 2537 (1994)

AI patent analysis frequently involves technical documents, software, diagrams and databases.

Copyright protection may apply to qualifying human-authored materials, including software and technical documentation.

However, patent protection and copyright protection are different.

A computer program may qualify for copyright protection while being excluded from patent protection as a computer program under Section 9(3) of the Patent Act.

3. AI Inventorship and Patent Ownership Disputes

DeepMind Launches Automated AI Research Lab in UK | The Tech Buzz

ArcPrime | AI-Native Patent Portfolio Optimization Platform

AI Tools Every Mechanical Engineer Should Know

10

One of the most important legal questions is whether an invention generated by an AI system can receive patent protection in Thailand.

Consider the following hypothetical example.

A Thai technology company develops an AI system that independently identifies a new industrial cooling mechanism. The system generates the technical design, identifies the materials and produces the operating instructions.

The company subsequently files a patent application naming the AI system as the inventor.

Several legal questions arise:

Can the AI system be recognised as an inventor?

Can the company claim ownership of the invention?

Must a human employee be identified as the inventor?

Does the company have a contractual right to the invention?

Can the invention satisfy the novelty and inventive-step requirements?

Under the existing Thai Patent Act, the stronger legal interpretation is that the inventor must be a natural person.

There is no established Thai judicial precedent recognising an AI system as an inventor.

However, this does not mean every AI-assisted invention is excluded from patent protection.

Where a human inventor makes the relevant inventive contribution and AI merely assists with research, modelling or technical development, a patent application may potentially qualify.

Human contribution versus AI assistance

4. Six Important Thai Judicial Decisions

8 relevant Thai patent decisions discussed

There is an important limitation: Thai courts have not yet established a body of decisions specifically concerning AI-generated patent inventions or AI patent-analysis governance.

The following decisions are genuine Thai patent-law precedents. Their principles are applied by analogy to AI-related disputes.

Case 1: Supreme Court Decision No. 9213/2557

Subject: Novelty and prior-art disclosure

The dispute concerned a petty patent for a fruit-wrapping bag.

The Court considered whether the invention had already been disclosed through foreign publications and existing products.

The Court invalidated the petty patent after considering the prior-art evidence.

Legal principles:

Novelty is assessed against relevant prior art.

Foreign publications can be relevant.

Prior disclosure must be established through reliable evidence.

Patent applicants cannot obtain protection for inventions that are already part of the state of the art.

Application to AI patent analysis:

AI patent-search software may identify similar inventions in international patent databases.

However, an AI-generated prior-art report must contain accurate publication details and verifiable technical comparisons.

A fabricated or incorrectly classified prior-art reference can undermine the reliability of patent examination.

Case 2: Supreme Court Decision No. 4131/2536

Subject: Inventive step and obviousness

The case involved a patented water-filtration machine consisting of multiple filtration tanks.

The Court examined the claimed technical characteristics against earlier foreign patents and generally known technical principles.

It concluded that the claimed technical arrangement did not establish a sufficient inventive step.

The patent was therefore not validly granted.

Legal principles:

An invention must involve more than an obvious technical arrangement.

Existing technical knowledge is relevant.

Minor technical modifications may fail the inventive-step requirement.

The claimed technical contribution must be examined objectively.

Application to AI patent analysis:

An AI system may produce a technically different arrangement by combining existing engineering concepts.

However, a novel-looking AI-generated configuration does not automatically constitute an inventive step.

Case 3: Supreme Court Decision No. 974/2551

Subject: Standing to challenge patent validity

This case concerned T.M. Grating Steel Co., Ltd. and Billion Mass Industry Co., Ltd.

The dispute involved a patented steel-grating product and whether the claimant had sufficient interest to challenge the patent.

The Supreme Court recognised that a party directly affected by the patent's existence could qualify as an interested person even though commercial production was still at an experimental stage.

Legal principles:

A person directly affected by a patent may have standing to challenge its validity.

Actual commercial production is not necessarily required in every situation.

The nature of the party's commercial interest matters.

Patent challenges require a legally recognised interest.

Application to AI patent analysis:

A company developing an AI-designed industrial product may need to challenge an existing patent that restricts its commercial activities.

The decision is relevant to determining whether such a company has sufficient legal standing.

Case 4: Supreme Court Decision No. 67/2566

Subject: Interested persons and invalidity of petty patents

This relatively recent decision considered the right to challenge the validity of a petty patent.

The Court distinguished between:

Raising invalidity as a defence in litigation.

Bringing an independent action seeking cancellation of a petty patent.

It explained that the right to bring an independent cancellation action is restricted to an interested person or the public prosecutor.

An interested person must have a direct relationship with and be affected by the existence of the patent.

Legal principles:

Patent validity challenges are subject to standing requirements.

Not everyone can bring an independent cancellation action.

A person accused of infringement may raise relevant validity arguments as a defence.

Direct commercial or legal impact is important.

Application to AI patent analysis:

An AI patent-monitoring platform may identify a potentially invalid competitor patent.

However, identifying a possible defect does not automatically give the platform or its owner standing to bring a cancellation action.

The legal interest of the actual claimant must be established.

Case 5: Supreme Court Decision No. 1384/2550

Subject: Technical differences and patent infringement

This decision concerned a dispute involving mechanical joints used for reinforcing steel bars in construction.

The Court considered whether the defendant's manufacturing process differed sufficiently from the patented technical process.

The technical differences were examined in light of the underlying characteristics and functions of the invention.

Legal principles:

Minor technical differences do not necessarily avoid infringement.

The actual technical characteristics must be examined.

A competing process must be compared with the protected invention.

Technical evidence is important in patent disputes.

Application to AI patent analysis:

AI-powered patent comparison systems may identify differences between two industrial processes.

However, an infringement determination must consider the scope of the patent claims and the legally relevant technical characteristics, rather than merely comparing surface-level similarities.

Case 6: Supreme Court Decision No. 11753–11754/2557

Subject: Confidential technical information and alleged copying of product specifications

The litigation involved container-liner products and allegations that technical product specifications and drawings had been copied following a commercial relationship.

The dispute illustrates the importance of protecting technical information and commercial documentation when businesses develop and manufacture competing products.

Legal principles:

Technical drawings and specifications may be important evidence in intellectual property disputes.

Prior commercial relationships can be relevant to disputes concerning access to technical information.

The source and use of confidential technical material must be examined.

A party must establish the basis of its claimed rights and alleged infringement.

Application to AI patent analysis:

If an employee uploads confidential engineering drawings into an external AI platform, the resulting dispute may concern unauthorised disclosure, misuse of trade secrets or contractual confidentiality obligations.

Case 7: Supreme Court Decision No. 873/2544

Subject: Patent infringement and preliminary judicial protection

The dispute involved Thai Nippon Concrete Co., Ltd. and a person accused of patent infringement.

The case concerned an application for interim protection in a patent dispute.

The Court considered the relationship between the asserted patent rights, the evidence of infringement and the requirements for judicial interim relief.

Legal principles:

Interim injunctions require appropriate evidentiary support.

A claimant must establish the legal basis for the asserted patent rights.

Patent validity and infringement are relevant to requests for urgent protection.

Courts must examine the circumstances before granting interim remedies.

Application to AI patent analysis:

An AI system may detect the unauthorised use of a patented industrial process.

A patent owner seeking an injunction must still establish the relevant patent rights and provide convincing evidence of infringement.

An automated infringement alert is not, by itself, sufficient proof.

Case 8: Supreme Court Decision No. 9832–9833/2560

Subject: Patent claim interpretation and infringement analysis

These proceedings are relevant to the judicial interpretation of foreign patent doctrines in Thai patent litigation.

The associated research examines doctrines including:

Claim construction.

Literal infringement.

Doctrine of equivalents.

Prosecution-history interpretation.

Obviousness.

Anticipation.

Legal principles:

Patent claims define the scope of protection.

Technical descriptions and prosecution materials can assist interpretation.

Infringement analysis requires legal and technical assessment.

Foreign patent doctrines must be applied consistently with Thai legislation.

Application to AI patent analysis:

AI patent-analysis systems can compare patent claims and identify potentially equivalent technical features.

However, their output must be assessed under Thai patent law rather than automatically importing the legal conclusions of foreign jurisdictions.

The decisions and their relevance are supported by Thai patent-law case materials and specialist analyses.

IPLINK ASIA+5

5. Major Categories of AI Patent Analysis Governance

A. AI-Based Patentability Assessment

How to complete a patentability search in IPRally in minutes using an image | IPRally Blog

iprally.com

AI systems evaluate whether an invention satisfies patentability requirements.

They can compare proposed inventions with:

Existing patent documents.

Scientific publications.

Technical databases.

Engineering designs.

Earlier commercial products.

Legal challenges:

Incorrect identification of prior art.

Inaccurate novelty assessments.

Unreliable inventive-step predictions.

Incomplete patent searches.

Incorrect patent classifications.

The final patentability decision remains governed by the Patent Act and the competent patent authorities.

B. AI Patent Ownership Governance

AI-assisted inventions can generate disputes between:

Employers and employees.

Universities and researchers.

Companies and software developers.

Investors and technology founders.

AI developers and commercial users.

For example, a university researcher develops a new pharmaceutical manufacturing process using an AI research platform.

The university, researcher and AI software provider may disagree about ownership.

Relevant legal considerations include Sections 10–12 of the Patent Act, employment agreements, research agreements and technology-transfer contracts.

C. AI Patent Infringement Detection

Suspected infringement analysis | IP Copilot

ipcopilot.ai

AI systems can compare patented products and processes with competing technologies.

Potential disputes include:

Unauthorised use of patented manufacturing processes.

Copying patented technical features.

Importation of infringing products.

Unauthorised commercial exploitation.

Disputes concerning equivalent technical features.

Section 36 and Section 36 bis are particularly important.

D. AI Patent Examination Governance

AI may assist the Department of Intellectual Property (DIP) in:

Patent classification.

Application screening.

Prior-art identification.

Document analysis.

Examination workflow management.

However, AI-assisted examination should preserve procedural fairness, accurate technical assessment and appropriate human responsibility.

E. AI Patent Portfolio Management

Companies increasingly use AI to analyse patent portfolios and identify commercial opportunities.

Potential disputes involve:

Incorrect valuation of patents.

Misleading investment reports.

Incorrect ownership records.

Undisclosed patent weaknesses.

Defective licensing recommendations.

Where these failures cause legally recognised financial damage, contractual or tortious liability may arise.

F. AI Patent and Trade Secret Conflicts

An invention may contain both patentable technical features and confidential commercial information.

A company must decide which information should be disclosed in a patent application and which information should remain protected as a trade secret.

Uncontrolled AI processing can expose confidential research to external service providers.

The Trade Secrets Act and contractual confidentiality protections are therefore important.

6. Civil Liability for AI Patent Analysis Errors

The Civil and Commercial Code provides the general legal foundation for claims arising from negligent AI patent analysis.

Example 1: Incorrect Patentability Report

A company pays an AI patent consultancy to evaluate a new industrial invention.

The consultancy's system incorrectly concludes that the invention is entirely novel.

The company invests substantial funds in manufacturing and marketing.

Later, the invention is found to have been publicly disclosed in an earlier patent publication.

Potential legal issues include:

Breach of the consultancy agreement.

Negligent professional services.

Failure to conduct appropriate verification.

Financial loss caused by inaccurate advice.

Section 420 may become relevant if the necessary elements of wrongful conduct and causation are established.

Example 2: AI Patent Infringement Misclassification

A company uses AI to examine a competitor's patented machinery.

The system incorrectly reports that the company's product does not infringe.

The company begins manufacturing and distributing the product.

The patent owner subsequently brings infringement proceedings.

The company may face commercial losses, litigation expenses and other consequences.

Potential liability may arise if the AI service provider breached contractual obligations or negligently performed its agreed services.

Example 3: Wrongful Disclosure of Patent Information

An employee uploads unpublished patent documents into a publicly accessible AI platform.

The documents contain confidential engineering designs.

The information becomes available to third parties.

Potential claims may arise under:

The Trade Secrets Act.

Employment contracts.

Confidentiality agreements.

Section 420 of the Civil and Commercial Code, where its requirements are satisfied.

7. AI Patent Analysis and Personal Data Protection

The Personal Data Protection Act B.E. 2562 may become relevant when AI patent platforms process identifiable personal information.

Examples include:

Inventor identification records.

Researcher employment information.

Patent ownership databases.

Biometric access records.

Confidential employee research histories.

Important provisions include:

Section 19: Consent.

Section 23: Information obligations.

Section 24: Lawful bases for processing.

Section 30: Data-subject access rights.

Section 35: Accuracy of personal data.

Section 37: Data-controller obligations.

Section 40: Data-processor obligations.

An AI patent-analysis company must consider whether its processing of inventor and researcher information complies with applicable PDPA requirements.

Patent documentation may contain commercially sensitive information, but not every confidential technical document constitutes personal data.

8. Institutional Governance and Regulatory Authorities

ประวัติกรมทรัพย์สินทางปัญญา

Department of Intellectual Property (DIP)

Patent applications, examination, registration and intellectual property administration.

ระบบติดตามและประเมินผลการส่งเสริม MSME v.0.0.1

Ministry of Commerce

Intellectual property policy and administration.

ศาลทรัพย์สินทางปัญญาและการค้าระหว่างประเทศกลาง : พร้อมไป (PromptPai)

Central Intellectual Property and International Trade Court

Specialised judicial forum for patent infringement and other intellectual property disputes.

NIA Logo PNG Vector (SVG) Free Download

National Innovation Agency

Innovation development and support for commercialisation of technology.

Other institutions, including the Personal Data Protection Committee and relevant research institutions, may also become involved depending on the dispute.

9. Evidence Required in AI Patent Litigation

AI-related patent disputes require technical as well as legal evidence.

A particularly important issue is the preservation of AI development history.

Businesses should maintain records identifying:

Who supplied the original technical idea.

Which AI tools were used.

What prompts and instructions were supplied.

What technical modifications were made by humans.

Who selected and validated the final invention.

Which contractual arrangements govern ownership.

These records may become critical when inventorship or ownership is disputed.

10. Comparative International Case Law on AI Inventorship

Thailand has not yet established a judicial position specifically recognising AI inventorship. Two foreign proceedings provide useful comparative guidance.

A. Thaler v Comptroller-General of Patents

United Kingdom Supreme Court, 2023

The dispute concerned patent applications identifying the AI system DABUS as the inventor.

The UK Supreme Court concluded that an inventor under the applicable UK patent legislation must be a natural person.

Relevance to Thailand: This reasoning is useful when interpreting the human-inventor framework under Thailand's Patent Act.

B. Thaler v Vidal

United States Court of Appeals for the Federal Circuit, 2022

The dispute concerned whether an AI system could be identified as the inventor of patent applications.

The Federal Circuit held that the US Patent Act requires an inventor to be a natural person.

Relevance to Thailand: The decision illustrates the difficulty of granting patent rights where no human inventor can be identified.

These are comparative authorities, not binding Thai precedents.

11. Recommendations for Reforming AI Patent Governance in Thailand

Thailand could consider the following reforms.

1. AI inventorship clarification: Establish statutory guidance concerning human contributions to AI-assisted inventions.

2. Ownership rules: Clarify ownership of inventions produced through employment, commissioned research and AI-assisted development.

3. AI patent examination: Introduce standards for the responsible use of AI in patent examination.

4. Technical audit requirements: Encourage independent verification of AI-generated patentability assessments.

5. Human accountability: Require meaningful human verification of important patent decisions.

6. Confidentiality protection: Establish stronger safeguards for unpublished patent applications and research data.

7. AI infringement assessment: Develop guidance for AI-assisted comparison of patent claims and technical features.

8. Research institution governance: Clarify ownership and remuneration arrangements for AI-assisted academic inventions.

9. Civil remedies: Improve clarity concerning compensation for losses resulting from negligent AI patent services.

10. International cooperation: Strengthen cooperation with WIPO and foreign patent offices on AI-generated inventions.

12. Conclusion

Civil Law and AI Patent Analysis Governance in Thailand is an emerging area involving patent ownership, inventorship, technological innovation, contractual responsibility and intellectual property enforcement.

The Patent Act B.E. 2522 remains the central legal framework. Sections 5–9 govern patentability and exclusions, Sections 10–12 address inventorship and ownership, and Sections 35–36 bis govern patent rights and their scope.

The eight Thai Supreme Court decisions discussed above provide relevant principles concerning novelty, inventive step, patent infringement, patent validity and standing to challenge patents.

However, none should be interpreted as directly deciding whether an AI system can qualify as an inventor.

Three fundamental legal conclusions

AI-assisted inventions may qualify for Thai patent protection if they satisfy the statutory requirements and the inventorship requirements are properly addressed.

AI systems themselves are not presently recognised under an established Thai legal framework as independent patent inventors.

Companies using AI for patent research, examination and infringement analysis remain responsible for complying with applicable legal, contractual and confidentiality obligations.

Academic research qualification: The discussion distinguishes binding Thai legislation, actual Thai judicial decisions and comparative foreign authorities. It does not suggest that Thailand already has an established AI-specific patent liability doctrine.