According to the Basic Act for Artificial Intelligence (“AI Act”) passed and announced in January 2026, the term “artificial intelligence” shall refer to a system capable of autonomous operation, which, through input or sensing and by means of machine learning and algorithms, can generate outputs such as predictions, content, recommendations, or decisions that affect physical or virtual environments to achieve explicit or implicit objectives.

Also, he “Guidelines for the Use of Artificial Intelligence (AI) in the Financial Industry” (“Financial Industry AI Guidelines”) issued in June 2024 by Taiwan’s financial regulator, the Financial Supervisory Commission (“FSC”) define “AI systems” as “systems that learn from large amounts of data and use machine learning or related model-building algorithms to perform perception, prediction, decision-making, planning, reasoning, communication, etc. to imitate human learning, thinking and reaction patterns”.

Taiwan government announced the “Taiwan AI Action Plan” in 2018 to declare Taiwan’s goal to become an important partner in the value chain of global AI technology and intelligence systems and to leverage the advantages of software and hardware techniques to promote AI technology across industries. According to the Taiwan AI Action Plan, the government’s view is that Taiwan is well positioned to take advantage of the opportunities in developing AI-related industries.

In 2023, the government unveiled the “Taiwan AI Action Plan 2.0,” building on earlier initiatives with a focus on five core objectives: expanding and optimizing talent, advancing technology and industrial development, enhancing operational environments, increasing international influence, and tackling human and societal issues. Moreover, in the inauguration speech of Taiwan’s newly elected President Lai on May 20, 2024, he emphasized the nation’s commitment to becoming an “Island of Artificial Intelligence” in response to global and technological challenges. This vision includes promoting the industrialization of AI, fast-tracking innovative AI applications, and incorporating AI across various sectors. The overarching goal is to harness AI computing power to boost national strength, military capabilities, human resources, and economic growth.

In addition, recent media reports have indicated that Taiwan’s Executive Yuan (cabinet) will soon launch the “New Ten Major AI Initiatives.” The “New Ten Major AI Initiatives” will encompass six major AI infrastructures and four advanced technologies:

(1) Infrastructure: The government will establish supercomputers and cloud data centers across Taiwan to expand computing capacity, support enterprises in big data processing and AI computation, and strengthen data governance and cybersecurity mechanisms to ensure the safe and stable application of technology. At the same time, the government will foster the AI platform software industry, aiming to assist one million enterprises in adopting AI technologies and promote the intelligent transformation of sectors such as transportation, healthcare, and education.

(2) Advanced Technologies: The policy will focus on four core areas: intelligent robotics, quantum technology, silicon photonics, and sovereign AI. Intelligent robotics applications will cover manufacturing and daily life, enhancing production efficiency. Quantum technology will open up a new computing architecture, potentially overcoming bottlenecks in semiconductors and optoelectronics. Silicon photonics is key to realizing high-performance AI chips, reducing energy consumption, and increasing speed. Sovereign AI will focus on cultivating domestic large language models and key technologies to ensure technological autonomy and independence. To this end, the government will form a “Quantum National Team” to integrate semiconductor and optoelectronic capabilities, solidifying Taiwan’s leading position in strategic AI technologies.

It is expected that by 2040, the “New Ten Major AI Initiatives” will generate approximately NT$15 trillion in economic output and create 500,000 high-paying jobs, injecting long-term momentum into industrial transformation and regional development.

Also, according to the AI Act, the Executive Yuan shall establish a National AI Strategic Committee, which shall coordinate, promote, and supervise national AI affairs and formulate the National AI Development Guidelines. The Committee shall convene at least once a year to review the National AI Development Guidelines. In the event of a sudden emergency or major incident, a special meeting shall be convened.

In 2019, the Ministry of Science and Technology under the Executive Yuan (the Cabinet) announced the “AI Technology R&D Guidelines” to demonstrate the Taiwan government’s commitment to improving Taiwan’s AI R&D environment. The government expects the participants to always be aware of the Guidelines when conducting relevant activities and endeavouring to build an AI-embedded society with three core values, which are “Human-centred Values”, “Sustainable Developments” and “Diversity and Inclusion”. Based on the above three core values, the following eight guidelines were published under the AI Technology R&D Guidelines for the guidance of AI participants, so that a solid AI R&D environment and society that connect to the global AI trends may be established: “Common Good and Well-being”, “Fairness and Non-discrimination”, “Autonomy and Control”, “Safety”, “Privacy and Data Governance”, “Transparency and Traceability”, “Explainability” and “Accountability and Communication”.

It is important to note that in August 2023, the Executive Yuan (Taiwan’s cabinet) released guidelines for the use of generative AI by the Executive Yuan and its subordinate government authorities and agencies.

Regarding the financial sector, Taiwan’s Financial Supervisory Commission (FSC) introduced the “Core Principles and Related Promotion Policies for the Use of Artificial Intelligence (AI) by the Financial Industry” in October 2023. Subsequently, on June 20, 2024, the FSC issued the “Guidelines for the Use of Artificial Intelligence (AI) in the Financial Industry” (“Financial Industry AI Guidelines”). These guidelines are intended to serve as a reference for financial institutions in the implementation, application, and oversight of AI technologies. The FSC has clarified that the Financial Industry AI Guidelines are intended as administrative guidance. Financial industry associations that have developed their own self-regulatory frameworks for AI usage can integrate the key points and measures from these Guidelines into their own regulations. In the absence of such self-regulations, financial institutions are advised to adhere to these Guidelines for the implementation, utilization, and management of AI systems.

As to the AI Act, although the Act does not impose concrete regulatory obligations on the private sector, it articulates several foundational principles intended to guide future AI governance:

1. Guiding Principles for AI R&D and Application: (1) sustainable development and well-being, (2) human autonomy, (3) privacy protection and data governance, (4) cybersecurity and safety, (5) transparency and explainability, (6) fairness and non-discrimination, and (7) accountability.

2. The government shall ensure that the application of AI does not result in any of the following: infringement upon the people’s life, body, liberty, or property; disruption of social order, national security, or the ecological environment; or conduct that violates relevant laws and regulations, including bias, discrimination, false advertising, or the dissemination of misleading or false information.

3. The government shall act in the best interests of children and adolescents. If an AI product or system is identified as a high-risk application by a central competent authority for the relevant industry in consultation with the Ministry of Digital Affairs (“MODA”), it shall come with advisory notices or warnings. The MODA and other relevant agencies shall provide or recommend assessment and verification tools or methods to assist the competent authorities for the relevant industries in carrying out such matters.

4. When developing, training, testing, and verifying the operational impact of emerging AI technologies, the government shall provide for reasonable use, support, and subsidy measures, and shall improve the legal framework for AI research, development, and application. In cases where the interpretation and application of relevant regulations conflict with other laws, the principle of promoting the provision of new technologies and services shall take priority.

5. To promote the innovation and sustainable development of AI technology, the competent authorities for the relevant industries may, for innovative AI products or services, establish or complete a safe environment for innovative AI R&D and application services experimentation.

6. The government shall establish mechanisms for open data, data sharing, and data reuse to enhance the availability of data for AI, and shall regularly review and revise relevant laws and regulations. The government shall strive to improve our quality and quantity of data used for AI, ensuring that training and output results sufficiently reflect the nation’s diverse cultural values and protect intellectual property rights. Each competent authority for the relevant industries shall consult with the competent authority for personal data protection, and shall, in the process of AI R&D and application, avoid the unnecessary collection, processing, or use of personal data, and shall promote the integration of personal data protection by design and by default measures or mechanisms to safeguard the rights and interests of the data subjects.

7. The MODA shall, with reference to international standards or norms, promote an AI risk taxonomy and assessment framework that is interoperable with international frameworks, and shall assist the competent authorities for the relevant industries in establishing risk-based management regulations. The competent authorities for the relevant industries shall, as needed for AI application risk management, follow the AI risk taxonomy and assessment framework to establish risk-based management regulations, and shall assist the relevant industries in formulating their own industry guidelines and codes of conduct.

8. For the application of high-risk AI, the government shall clarify the attribution of liability and the conditions for such liability, and shall establish relief, compensation, or insurance mechanisms.

Based on the passage of the aforementioned AI Act, the government/relevant agencies will draft more detailed regulations. Stakeholders are advised to closely monitor the upcoming regulatory developments.

As indicated in Question 3, the AI Act does not impose concrete regulatory obligations on the private sector, so currently there are no manadatory requirements for artificial intelligence transparency, explainability, or audits, nor mandatory obligations to disclose the use of AI to customers/clients.

However, as more detailed/subordinate regulations will be announced under the AI Act, it is expected that there would be such types of mandatory requirements or obligations in the future, especially with respect to relatively highly regulated industries (e.g., financial, medical, etc.)

As indicated in Question 3, the AI Act does not impose concrete regulatory obligations on the private sector, so currently there are no manadatory requirements for human oversight and human-in-the-loop in artificial intelligence systems. However, as more detailed/subordinate regulations will be announced under the AI Act, it is expected that there would be such types of mandatory requirements in the future, especially with respect to relatively highly regulated industries (e.g., financial, medical, etc.)

As to best practice expectations, these should primarily revert to the fiduciary duty of the company’s directors and senior management, as well as the duty of care exercised by individual personnel when using AI. Specifically, for example, before a company procures an AI solution, it should engage internal or external professionals to conduct adequate due diligence. Furthermore, from a systemic/mechanism perspective, if there is no certainty that the risks associated with AI-related products, services, or internal operations are reasonably acceptable (since the risk may never be zero), then in theory, there should still be room for human intervention. Furthermore, in terms of system design, individual personnel should also be assigned supervisory and control responsibilities.

As indicated in Question 3, the AI Act does not impose concrete regulatory obligations on the private sector, so currently there are no manadatory legal or regulatory requirements specifically addressing algorithmic bias, discrimination, or fairness in AI systems (including gender bias). However, as more detailed/subordinate regulations will be announced under the AI Act, it is expected that there would be such types of mandatory requirements in the future, especially with respect to relatively highly regulated industries (e.g., financial, medical, etc.)

Civil liability:

As AI has not yet been recognised as a legal entity under current Taiwan law, it cannot be liable for any civil liability under current Taiwan law. Therefore, in case of tort liability arising from the use of AI technology, the injured party would still need to prove that the torts fall within any of the specific types of tort under the Civil Code and/or the Consumer Protection Act (“CPA”).

For example, the manufacturer of a self-driving car may be held liable under Article 7 of the CPA if the court considers that it is unable to prove that the car has met and complied with the contemporary technical and professional standards of reasonably expected safety requirements before such car was released on to the market. The injured person may also wish to prove and convince the judge that the self-driving vehicle falls within the definition of “automobile” and the user should be considered the “driver” under Article 191–2 of the Civil Code. In addition, the injured person would also bear the burden to prove that the “user” was negligent when using the self-driving vehicle if such person wishes to establish a claim under Article 184 of the Civil Code (i.e., the general tort).

The Taiwan court recently rendered a decision holding a driver fully liable for damages under Articles 184 and 191-2 of the Civil Code, for the driver’s reliance on the Advanced Driver Assistance System (ADAS)—an artificial intelligence feature embedded in the vehicle—and, as a result, failed to maintain adequate vigilance regarding road conditions, leading to a collision with a highway facility. The court determined that the driver’s dependence on ADAS did not absolve them of their legal duty to exercise due care, and thus imposed full civil liability for the resulting damages.

Criminal liability:

Criminal liability under Taiwan law typically requires a person’s mental state to allow “intention” or “negligence”. AI itself would not be capable of acquiring the aforementioned “mental state” and therefore of committing a criminal offence. In addition, in principle, only natural persons are deemed capable of committing crimes, save for certain exceptional circumstances where legal persons may be subject to criminal fines. Given that, similarly to the discussion on tort liability, with regard to the issue of determining whether a criminal offence has been committed, one would need to prove the required conditions of criminal liability, such as “intention” or “negligence” and “causation” on the part of the person “using” or “behind” the AI.

As indicated in Question 3, the AI Act does not impose concrete regulatory obligations on the private sector, so currently there are no mandatory cybersecurity obligations specifically apply to AI systems under the AI Act. However, as more detailed/subordinate regulations will be announced under the AI Act, it is expected that there would be such types of mandatory in the future.

On a separate note, Taiwan’s Cyber Security Management Act (“CSMA”) was amended in September 2025. Before the amendment, government agencies are generally prohibited from downloading, installing, or using products that pose risks to national cyber security (“Risky Products”) under the “Principles for Restricting the Use of Risky Products by Government Agencies.” This amendment elevates said principles to the level of CSMA. With the development of AI-related products and services, many new AI-related products are being introduced, and the aforementioned Risky Products may be related to AI.

At this stage, we have not seen any news about any insurance product specifically covering AI liability or liability stemming from the use of AI. However, the views of many experts and scholars indicate that the insurers should strive to design suitable AI liability insurance products in collaboration with car manufacturers and software companies (in the context of insurance for liability caused by self-driving cars) as well as the hospitals, medical doctors and software companies (in the context of insurance for liability caused by medical AI). There are also articles by legal scholars which analyze the possible mechanisms for covering and allocating the risks and liabilities caused by AI, such as liability insurance or special compensation funds.

In judicial practice, an artificial intelligence device cannot be named as an inventor of a patent. Judgments from the Taiwan Intellectual Property and Commercial Court hold that a patent invention is the creative output of the human spirit, and cannot be created by an artificial intelligence device; from the perspective of Taiwan law, only natural or legal persons can enjoy such rights.

Determining the owner of the intellectual property in an AI-created work is expected to be a legal issue that will be widely discussed as the use of AI becomes more widespread. According to the views of many experts and scholars, currently AI learns through computer software designed by humans, which is called “deep learning”. In addition to responding to human query inputs, AI is able to use its limited intrinsic perception and logic to help its users make decisions. In other words, as AI already has the ability of deep learning, it is not merely a tool for humans. However, there would be issues as to whether AI has the ability to create an “original expression” under copyright law. In this regard, we believe that as of now, as AI is still not a “person” from a legal perspective, work created by AI cannot be copyrighted — such view is also generally supported by a letter of interpretation issued by Taiwan’s Intellectual Property Office (IPO) dated April 20, 2018 (Ref. No.: 1070420).

In general, our preliminary view is that such issues might not be resolved under the current IP regime in Taiwan; it is a real challenge faced by, and needs to be addressed by, the government, legislators, representatives of the court system, and other legal practitioners in the future along with the development of AI.

Two trending issues with AI tools:

(1) Confidentiality: As AI tools collected data from the users, it is generally suggested that confidential or sensitive information (e.g., trade secret) should be removed or “de-identified” before using the AI tools.

(2) Correctness: It is still the general view that content (especially facts) provided by relevant AI tools might not necessarily be factually accurate, and the information generated by AI tools would need to be verified independently in order not to make mistakes in the workplace due to the use of inaccurate information generated by AI tools.

With respect to employment, the AI Act provides that the government shall actively use AI to safeguard workers’ labor rights, and shall actively bridge the skills gap caused by AI development, enhance labor participation, ensure economic security, and implement dignified labor. Also, the government shall provide employment counseling to those who become unemployed due to the use of AI, based on their work capabilities. However, as indicated in Question 3, the AI Act does not impose concrete regulatory obligations on the private sector, so currently there are no specific laws or regulations to address issues arising from AI-driven hiring, performance evaluation, or employee monitoring. However, we anticipate that this topic will increasingly become a subject of discussion within the field of labor and employment law. As such, it is likely to develop further in both legal practice and court proceedings should any disputes arise in the future.

In Taiwan, personal data is protected by Taiwan’s Personal Data Protection Act (“PDPA”). The collection, processing and use of any personal data are generally subject to notice and consent requirements under the PDPA. Pursuant to the PDPA, personal data is defined broadly as the name, date of birth, ID card number, passport number, characteristics, fingerprints, marital status, family information, education, occupation, medical record, medical treatment and health examination information, genetic information, information about sex life, criminal record, contact information, financial conditions, social activities and other information that may directly or indirectly identify an individual.

Under the PDPA, unless otherwise specified by law, a company is generally required to give notice to (notice requirement) and obtain consent from (consent requirement) an individual before collecting, processing or using any of said individual’s personal data, subject to certain exemptions. To satisfy the notice requirement, certain matters must be communicated to the individual, such as the purposes for which his or her data is collected, the type of the personal data and the term, area and persons authorised to use the data, among other things.

In addition, “sensitive personal data” (i.e., personal data pertaining to a natural person’s medical records, healthcare, genetics, sex life, physical examination, and criminal records) would be subject to stricter regulations under the PDPA. For example, the consent must be made in writing, and the following must be complied with: (i) the collection, processing or use must not exceed the necessary scope of the specific purpose(s); (ii) the collection, processing or use based solely on the consent of the data subject is not otherwise prohibited by law; and (iii) such consent is not given by the data subject out of his/her free will

AI technology has not changed said requirements. If a company wishes to collect, process and use any individual’s personal data with respect to development (including training) and using AI technology or exploring the data with AI technology, it will be subject to the obligations under the PDPA as advised above.

There are no laws and regulations specifically promulgated or amended for regulating data scraping in relation to IP, privacy and competition in Taiwan. One needs to look into the current IP laws, the Personal Data Protection Act (“PDPA”), and the Fair Trade Act (“FTA”) to examine any legal issue that may arise.

In the field of IP, for example, if data scraping would involve the “reproduction” of copyrighted work of another, then it might constitute copyright infringement.

Regarding privacy, it is less likely that any data scraping activities would constitute violation of the PDPA if such activities only capture data generally available on the Internet.

In terms of competition law, Article 25 of the FTA prohibits enterprises from having any deceptive or obviously unfair conduct that can affect trading order. There was a case where an entity in the real estate industry scrapped data from a competitor’s website/App and made such scrapped data available at such entity’s own App. This act was found to have violated Article 25 of the FTA by the competent authority and the Taiwan Intellectual Property and Commercial Court.

In addition, it is worth noting that there was a court judgment in June 2025, in which the court held that a legal database service provider, by obtaining another provider’s data through data scraping, constituted a violation of the Criminal Code provision regarding the criminal offense of “interfering with computer use”, namely “without reason, obtains the magnetic record of another’s computer or related equipment and causes injury to the public or others”. However, if if the information was not obtained by unauthorized access to another party’s computer system, whether this should constitute the element of “without reason” remains to be observed in subsequent appeal proceedings.

From a Taiwan legal viewpoint, the prohibition of data scraping in the terms of use of a website may be enforceable if there is evidence that may prove that the visitor has been clearly inform of the terms of use (including such prohibition) and given consent to such terms of use (e.g, clicking the bottom “I agree” in order to continue to browse the website).

According to the AI Act, the competent authority under the Act refers to the “National Science and Technology Council” at the central government level, and the special municipality, county, or city governments at the local level.

However, as advised in Questions 3 above, certain government authorities are tasked with certain obligations with respect to AI. For example:

• if an AI product or system is identified as a high-risk application by a central competent authority for the relevant industry in consultation with the MODA, the government shall come with advisory notices or warnings. The MODA and other relevant agencies shall provide or recommend assessment and verification tools or methods to assist the competent authorities for the relevant industries in carrying out such matters.
• the MODA shall promote an AI risk taxonomy and assessment framework that is interoperable with international frameworks, and shall assist the competent authorities for the relevant industries in establishing risk-based management regulations.

In Taiwan, the use of AI has been more and more widespread, and the use cases can be found in industries such as transportation, healthcare, financial services, retail, media, etc. For example:

In transportation, the sandbox law for autonomous and self-driving vehicles, the Unmanned Vehicle Technology Innovation and Experiment Act (the Unmanned Vehicle Sandbox Act), provides a friendlier environment to test the applications of AI and the Internet of Things (IOT) in transportation. As of 2 July 2024, almost 20 innovative experimentations have been approved to enter the sandbox, and the commercial launch of them is generally expected as long as the results of the experiments are positive.

As to financial services, robo-adviser services (i.e., online securities investment consulting services using automation tools and algorithms) evolve with the promulgation and the subsequent amendments to the “Operating Rules for Securities Investment Consulting Enterprises Using Automated Tools to Provide Consulting Service”. In addition, in May 2025, Taiwan’s FSC released the results of a survey on the application of AI by financial institutions. The highest adoption rate of AI was found in banks (87%), followed by life insurance companies (67%) and property insurance companies (45%). The survey showed that the main purposes for financial institutions to apply AI are to improve operational efficiency (30%), save manpower (18%), and optimize customer experience (15%). The main areas of application are internal administrative operations, intelligent customer service, and the prevention of financial crimes. The most commonly used AI technologies are natural language processing (NLP) and large language models (LLM) (31%), machine learning (28%), and robotic process automation (RPA) (23%). For the financial industry, the most prevalent AI application areas remain internal administrative operations (39%) and intelligent customer service (15%).

In healthcare, recently, AI’s involvement in healthcare has broadened to include medical management, diagnostic processes, predictive diagnostics, and clinical care. For instance, major medical research institutions in Taiwan have been developing AI algorithms to be used for cardiovascular risk assessment, diagnosing cancer lesions at an early stage, accelerating the image recognition, among other things.

It is not surprising if any individual Taiwan lawyers or in-house counsels use AI tools for their work. Also, according to our understanding, there are law firms that have set, or are discussing whether to set internal policies governing lawyers’/employees’ use of AI tools in the office, and the rules or restrictions on the types of work products that may involve the assistance of AI tools, as well as relevant restrictions to protect client confidentiality. Due to the relatively smaller scale of law firms in Taiwan compared to those in Europe, the United States, and China, it is less likely for them to develop their own AI systems. However, most firms would at least consider using third-party AI—either legal tech solutions specifically designed for the legal industry or general-purpose AI software tools.

In general, the use of AI raises concerns regarding confidentiality (such as the risk of clients’ confidential information being collected or leaked), the accuracy of AI-generated content, and potential infringement of others’ rights (for example, copyrights).

The following are certain challenges raised by AI that have been discussed by lawyers in Taiwan: (1) types of work at risk of being replaced by AI (i.e., replacement of legal-related human resources); (2) clients’ unwillingness to pay for certain services that can be invoiced traditionally; (3) less opportunities for junior lawyers to get trained as their predecessors before the AI era; (4) the need to spend more time and energy to acquire the skills and knowledge needed for using AI; and (5) potential risk of being surpassed by competitors who are better at utilizing AI technologies.

Lawyers in Taiwan also have discussed the following potential opportunities that may arise due to the development of AI in private practice: (1) lawyers can be released from the more trivial tasks to higher-value work; (2) increase efficiency in completing the tasks; (3) provide services with more objective foundations and analysis to customers; (4) automate tasks; (5) potential feasibility to move from human-rendered services to services without human intervention (e.g., software, AI as legal services developed by the firm, etc.).

Please see our responses to above Question 3 regarding the AI Act. The regulatory development should be monitored closely as it progresses through the process of relevant rule-making.