Taiwan drafts AI Basic Act amid chip dominance and regulatory gaps

Taiwan superior its synthetic intelligence regulatory framework on August 28, 2025, when the Govt Yuan submitted the Draft Synthetic Intelligence Primary Act to the Legislative Yuan for deliberation. The proposed laws represents the end result of efforts courting again to 2017, when the island’s Ministry of Science and Know-how first launched the AI Grand Technique for a Small Nation undertaking to put money into creating an AI ecosystem.

The Draft AI Primary Act designates the Ministry of Digital Affairs because the competent authority liable for implementing the framework. The laws establishes elementary rules for presidency motion in advancing AI analysis and improvement, setting coverage goals primarily based on what the doc describes as “AI elementary rules and the federal government’s priorities.”

Taiwan’s strategic place within the international {hardware} provide chain has grown considerably because the introduction of generative AI. The introduction of techniques like OpenAI’s ChatGPT 3.0 in late 2022, together with different rising AI functions, strengthened Taiwan’s function in offering important {hardware} infrastructure for computational processing energy within the AI business. The semiconductor business’s manufacturing worth, primarily pushed by the semiconductor foundry sector, is projected to exceed NT$6.4 trillion in 2025, reflecting a 22.2 p.c improve from 2024, in accordance with the Industrial Know-how Analysis Institute.

The federal government’s method builds on earlier motion plans. Taiwan AI Motion Plan 1.0 ran from 2018 by way of 2021, specializing in analysis and improvement, infrastructure, and expertise cultivation to facilitate software-hardware integration and edge AI whereas supporting development of Taiwan’s AI and robotics industries. In 2018, the Govt Yuan rolled out this preliminary program to boost Taiwan’s AI capabilities and foster improvement of tailor-made AI companies.

After attaining the Taiwan AI Motion Plan 1.0 milestones, the Govt Yuan permitted Taiwan AI Motion Plan 2.0 in 2023, aiming to raise Taiwan’s AI business worth past NT$250 billion by 2026. This second motion plan focuses on expertise improvement, business development, and international technological affect, reinforcing Taiwan’s place as a expertise innovation hub.

With the rise of generative AI in late 2022 and its fast improvement in 2023, the Nationwide Science and Know-how Council initiated packages together with the Reliable AI Dialogue Engine, developed by TAIWANIA II. The Reliable AI Dialogue Engine undertaking marks a milestone below the motion plan, alongside the Chip-Pushed Taiwan Trade Innovation Program, aiming to safe Taiwan’s main place in chip manufacturing and promote integration of superior silicon chips with frontier applied sciences.

The Legislative Yuan has mentioned numerous payments to deal with dangers and issues related to AI improvement since 2019. Nonetheless, all of those payments qualify as what the doc phrases “fundamental acts,” offering solely high-level rules and moral issues for AI improvement. In August 2025, legislators from totally different political events proposed greater than 10 variations of the Synthetic Intelligence Primary Act and carried out a consolidated overview of some variations. The Govt Yuan permitted its personal model of the Draft AI Primary Act and submitted it to the Legislative Yuan for additional motion.

The Draft AI Primary Act outlines 15 coverage goals for presidency implementation. The federal government should adequately plan general useful resource allocation and deal with issues regarding subsidies, commissions, investments, incentives, help and steering for AI-related industries or present monetary incentives resembling taxes or financing. Authorities companies shall overview and regulate their capabilities and operations, enacting, amending, or abolishing related rules or tips. When present rules stay silent previous to such enactment, modification or abolition, the federal government interprets and applies the Act in accordance with AI elementary rules by prioritizing promotion of recent applied sciences and companies.

Sectoral regulators could set up or full present progressive experimental environments for AI progressive services or products. The federal government ought to cooperate with the non-public sector to advertise AI innovation and utility by way of public-private partnerships. The federal government shall endeavour to advertise AI-related worldwide cooperation and take part in worldwide joint improvement and analysis programmes.

The laws mandates continued promotion of AI training in any respect college ranges, inside industries and throughout society, authorities companies and public establishments. The federal government should forestall AI functions from inflicting private or property injury, destruction of social order or the surroundings, battle of curiosity, bias, discrimination, false promoting, misinformation, falsification or different issues that violate related rules.

The Ministry of Digital Affairs and different related authorities could present or suggest evaluation and verification instruments for sectoral regulators to deal with these issues. The Ministry of Digital Affairs shall promote a danger classification framework for AI that aligns with worldwide requirements in order that sectoral regulators could stipulate their danger classification rules for the industries they oversee.

Authorities companies could set up requirements, verification, traceability or accountability mechanisms by way of legal guidelines or tips, adopting a danger administration method to evaluate potential vulnerabilities and abuses to boost the verifiability and human management of AI decision-making, and the trustworthiness of AI functions. For prime-risk AI functions, the federal government shall clearly outline the attribution of accountability and the circumstances for legal responsibility, establishing mechanisms for treatment, compensation or insurance coverage. AI techniques that stay within the analysis and improvement stage shall be exempt from these accountability necessities and mechanisms. Nonetheless, this exemption doesn’t apply if the AI is examined in real-world environments, or if the outcomes of AI R&D are used to supply services or products.

The federal government shall forestall ability gaps, shield labour rights and supply employment measures for many who will turn out to be unemployed attributable to using AI primarily based on their work capabilities. The competent authority for private information safety shall help sectoral regulators to advertise information safety by default and by design measures or mechanisms. The federal government shall set up a mechanism for open information, information sharing and information reuse to boost the supply of information utilized by AI and usually overview and regulate or adapt related legal guidelines and rules.

The federal government should endeavour to enhance the standard and amount of information utilized by AI and make sure that the coaching outcomes uphold nationwide multicultural values and respect mental property rights. When utilizing AI for its operations or provision of companies, the federal government shall conduct danger assessments and plan danger response measures. The federal government shall formulate guidelines of use or an inside management and administration mechanism in accordance with the character of the operations involving AI use.

Like different fundamental acts in Taiwan, the Draft AI Primary Act solely outlines the federal government’s high-level duties for fostering AI improvement and doesn’t impose particular regulatory necessities or obligatory obligations on the non-public sector. Nonetheless, as soon as the Draft AI Primary Act is enacted and applied, ministries and commissions are required to overview and regulate their legal guidelines and rules in accordance with the Act’s provisions.

Earlier than enacting or amending related legal guidelines and rules, present legal guidelines and rules have to be interpreted and utilized in keeping with the Draft AI Primary Act provisions. Monitoring the Draft AI Primary Act developments and observing how sectoral regulators will adapt their legal guidelines and rules after the Act’s enactment stays essential.

The AI Know-how R&D Tips, issued by the Nationwide Science and Know-how Council in September 2019, purpose to advertise moral AI analysis throughout numerous industries. These Tips spotlight three core values for AI R&D: human-centred values, sustainable improvement, and variety and inclusivity. To assist these values, the AI R&D Tips define eight rules for AI R&D, together with equity and non-discrimination, human autonomy and management, security, privateness and information governance, transparency and traceability, explainability, and accountability.

As a non-binding, voluntary framework, there are not any penalties for non-compliance, neither is there an enforcement physique. Nonetheless, their emphasis on moral AI analysis is essential for the accountable improvement and deployment of AI expertise.

The Public Sector Gen AI Tips, formulated by the Nationwide Science and Know-how Council and issued by the Govt Yuan in August 2023, set up a framework for the moral utility of generative AI inside the public sector. These tips emphasise the significance for presidency personnel to evaluate the knowledge generated by generative AI to make sure objectivity and professionalism, quite than changing their impartial judgment and creativity. Additionally they prohibit the dissemination of categorised or private information with AI techniques and suggest that generative AI shouldn’t function the only real foundation for administrative acts or public choices.

In June 2024, the Monetary Supervisory Fee issued non-binding administrative recommendation within the type of the Monetary AI Tips. These Tips purpose to ascertain finest practices for monetary establishments utilising AI and introduce a risk-based analysis framework to make sure accountable adoption of AI. The six core rules upheld by the Monetary AI Tips embrace governance and accountability mechanisms, equity and human-centred values, privateness and buyer rights safety, system robustness and safety, transparency and interpretability, and sustainable improvement.

Monetary establishments are required to conduct danger assessments and implement acceptable danger management measures when utilizing AI techniques, contemplating components such because the utilization of AI techniques, the extent of autonomy, complexity, impression on stakeholders and prospects for searching for reduction. The Monetary AI Tips apply to monetary establishments resembling banks, insurance coverage corporations and securities corporations, emphasising compliance with AI rules all through all the life cycle of AI techniques. Though the rules are non-binding, it’s anticipated that the Monetary Supervisory Fee will oversee compliance with them as a result of extremely regulated nature of the monetary sector.

The Ministry of Digital Affairs launched the Draft AI Analysis Tips in March 2024. Referencing the European Union’s Synthetic Intelligence Act, the Draft AI Analysis Tips classify AI merchandise and techniques into 4 danger ranges: unacceptable, excessive, restricted and low. They supply particular standards for evaluating AI merchandise, specializing in general high quality, efficiency and the flexibility to take care of performance below numerous circumstances. Focused on the data expertise, telecommunication, communication, cybersecurity and web sectors, these Tips encourage industries to self-monitor their AI merchandise or techniques primarily based on danger ranges and suggest submitting high-risk merchandise to an AI testing centre for analysis.

As a part of the AI Motion Plan 2.0, the Ministry of Digital Affairs additionally established the Synthetic Intelligence Analysis Middle to implement the analysis mechanism for AI merchandise and techniques. Whereas these Tips should not obligatory, they supply important steering and advised processes for business reference. The precise timeline for when they’ll take impact stays unsure.

The significance of equity and non-discrimination is highlighted in numerous AI tips and the Draft AI Primary Act. The AI R&D Tips, the Monetary AI Tips and the Draft AI Analysis Tips all emphasise the necessity to keep away from bias and discrimination within the improvement and deployment of AI techniques. The AI R&D Tips encourage the institution of exterior suggestions mechanisms to make sure equity, whereas the Monetary AI Tips require using numerous, high-quality information and third-party assessments to forestall bias.

The Monetary AI Tips additionally introduce the ideas of human-in-command, human-in-the-loop and human-over-the-loop to make sure equity. The Draft AI Analysis Tips prioritise equity as a vital analysis criterion, requiring AI techniques to deal with all teams equitably, regardless of race, gender, political opinions, disabilities and different components.

The Draft AI Primary Act equally establishes equity and non-discrimination as one of many elementary rules, urging AI builders and repair suppliers to keep away from algorithmic biases and discrimination. Moreover, it specifies that the Ministry of Digital Affairs and different related authorities could present or suggest instruments and strategies for evaluation and verification to forestall sure harms attributable to AI, together with bias and discrimination.

The Draft AI Analysis Tips set forth rigorous testing necessities to make sure the accuracy and reliability of AI techniques, significantly for high-risk AI techniques. These Tips define a complete four-step course of generally known as TEVV, encompassing a variety of things, together with security, explainability, resilience, equity, accuracy, transparency, accountability, reliability, privateness and safety. Equally, the Monetary AI Tips stipulate that monetary AI techniques should endure analysis for robustness and efficiency all through their total life cycle. This contains assessing their capacity to satisfy implicit or express goals through the design section and guaranteeing their efficient operation below real-world circumstances earlier than deployment. These Tips emphasise the need of thorough testing and analysis to make sure AI techniques’ effectiveness and reliability throughout numerous domains.

Within the realm of AI governance, transparency and accountability are essential rules which are typically upheld. The Public Sector Gen AI Tips and Monetary AI Tips each emphasise the significance of transparency by requiring acceptable disclosure of AI utilization in governmental and monetary establishments. Moreover, the Draft AI Analysis Tips sort out the problem of inadequate transparency in sure AI techniques and stress the need for explainability, enabling stakeholders to know the reasoning behind AI choices.

In the course of the R&D section, the AI R&D Tips recommend {that a} minimal degree of data be offered and disclosed when creating and making use of AI techniques, software program and algorithms, together with modules, mechanisms, parts, parameters and calculations. The eye to transparency ensures that individuals can perceive the weather concerned in AI techniques’ decision-making processes. AI improvement and utility ought to adhere to traceability necessities, resembling information assortment, information labelling and monitoring algorithms utilized in decision-making. Concerning accountability, these Tips for each sectors require the institution of inside management rules and acceptable governance constructions to make sure that authorities companies and monetary establishments assume accountability for any opposed outcomes ensuing from their AI techniques.

The Draft AI Primary Act equally emphasises transparency and explainability, in addition to accountability, as two elementary rules for AI analysis, improvement and utility. It recommends that the AI-generated content material be clearly disclosed or labelled to facilitate the evaluation of potential dangers and understanding of their impression on stakeholders. It additionally requires guaranteeing acceptable accountability, together with each inside governance and exterior social duties.

The applying of AI expertise, significantly generative AI, has profoundly impacted the content material business. The authorized points related to utilizing generative AI embrace the necessity for authorised information throughout AI coaching, the appliance of truthful use, the safety of generated outcomes below copyright legislation and the danger of infringement when utilizing AI-generated content material. Moreover, using generative AI could contain authorized points associated to infringement on portrait rights, voice rights and different persona rights. Taiwan has not but established particular legal guidelines or rules governing the appliance of AI expertise for content material technology. These points are typically addressed by present authorized frameworks, such mental property legal guidelines and the Civil Code. Nonetheless, owing to the rising nature of those authorized points and the shortage of adequate precedents, vital uncertainty stays in addressing potential dangers.

Notably, the Taiwan Mental Property Workplace has issued a number of rulings relating to copyright points in using generative AI. For instance, one ruling emphasises the necessity for copyright homeowners to authorise using their work for AI coaching. One other ruling implies the attainable infringement of copyrights when utilizing articles generated by ChatGPT and suggests first clarifying whether or not ChatGPT has obtained authorisation from copyright homeowners to keep away from disputes. The Taiwan Mental Property Workplace has additionally burdened that solely the creation by a pure individual may be protected by the Copyright Act. AI-generated content material will not be copyrighted if it doesn’t contain human artistic enter.

These rulings typically point out that if an AI mannequin makes use of work from any third events for coaching or imitation in AI techniques, it might represent copy of these works and should probably infringe upon the third events’ copyrights. The AI operator ought to safe authorisation or a licence from the copyright homeowners for using their work in AI coaching to forestall such infringement. Moreover, if the technology of content material doesn’t contain human effort, the output by AI won’t be protected by the Copyright Act. These rulings could present some steering when addressing mental property points within the context of AI.

At the moment, there is no such thing as a particular laws that addresses liabilities arising from using AI. Liabilities related to AI utilization are ruled by present regulatory frameworks, such because the Civil Code and Client Safety Act. Liabilities below the Civil Code typically observe the precept of fault legal responsibility, whereas liabilities below the CPA are strict liabilities. Nonetheless, establishing accountability for AI-related hurt below theories of negligence or product legal responsibility is difficult attributable to difficulties in proving causation and fault amongst customers, homeowners or builders of AI techniques. The absence of business requirements for AI merchandise additional complicates the willpower of legal responsibility below client safety legal guidelines.

Moreover, using AI in sectors resembling unmanned automobiles and medical therapy introduces challenges in figuring out legal responsibility. Within the realm of unmanned automobiles, the differing ranges of automation and management between AI expertise and human operators complicate the attribution of accountability within the occasion of an accident. Equally, inside AI-related medical disputes, the black-box nature of AI algorithms and the complexity of the decision-making processes current challenges in figuring out the legal responsibility of medical professionals and AI techniques, elevating points associated to the requirements of care anticipated from medical professionals and difficult their discretion.

Notably, the Draft AI Primary Act requires the Taiwan authorities to ascertain rules in regards to the circumstances, duties, treatments, compensation and insurances for AI functions primarily based on AI danger classification, aiming to make clear legal responsibility attribution and standards, and thereby enhancing the trustworthiness of AI.

The Draft AI Primary Act requires the federal government to forestall AI functions from inflicting hurt to residents’ lives or the surroundings, and to keep away from points resembling conflicts of curiosity, bias, discrimination, false promoting, misinformation or fraud. It additionally encourages the Ministry of Digital Affairs and related authorities to supply instruments and strategies for assessing and verifying AI to forestall such harms. The Statute of Fraud Crime Hurt Prevention, enacted by the Taiwan authorities in July 2024, additionally mandates internet marketing platforms to reveal using deepfake expertise or AI-generated private photos in commercials.

The disclosure and notice-of-use necessities purpose to boost transparency, accountability and client safety, in addition to to forestall hurt attributable to fraud. The Draft AI Primary Act emphasises the significance of transparency and explainability in AI techniques, requiring acceptable disclosure or labelling of AI outputs to guage dangers and perceive their impression on rights and pursuits. The Fraud Prevention Statute additionally mandates internet marketing platforms to reveal using deepfake expertise or AI-generated private photos in commercials.

The Draft AI Primary Act and related AI tips don’t clearly outline their jurisdiction, leaving uncertainty about whether or not the following AI rules have any extraterritorial impact. It stays unspecified how these rules will apply to AI merchandise from different nations.

The interplay between AI and competitors legislation poses challenges inside the digital economic system. In gentle of this, the Truthful Commerce Fee, in its White Paper on Competitors Coverage within the Digital Economic system issued in December 2023, highlighted issues about information assortment resulting in elevated value discrimination and the difficulties in investigating concerted actions by enterprises. Moreover, present rules in Taiwan haven’t adequately addressed the problems of differential therapy between enterprises and customers within the context of AI-driven personalised pricing.

In April 2024, the Truthful Commerce Fee fined Agoda for utilizing a competitor’s enterprise identify as a key phrase when deploying a self-learning machine to buy key phrase commercials, which prompted confusion amongst web customers. This case additionally highlights the potential challenges of utilizing AI within the realm of competitors legislation.

In 2025, the Truthful Commerce Fee printed the Explanatory Info on Soliciting Public Opinions on Competitors Regulation Points Associated to Generative Synthetic Intelligence in Taiwan to collect public feedback for the Truthful Commerce Fee’s future legislative and enforcement initiatives. This doc clarifies the market construction and traits of the generative AI, gives an outline of Taiwan’s AI {hardware} provide chain, mannequin improvement, and utility deployment, and additional addresses competitors points which will come up from generative AI. Particularly, it examines 4 varieties of regulated competitors conduct: unilateral abuse of market energy, concerted actions, mergers and false promoting or different unfair competitors practices.

The rules of information safety is without doubt one of the core rules upheld by the Monetary AI Tips. When utilizing AI techniques to supply monetary companies and work together with clients, monetary establishments ought to shield buyer privateness by adhering to the precept of information minimisation and guarantee transparency by informing clients about how their rights and advantages could also be affected.

The Draft AI Primary Act additionally enshrines the precept of privateness safety and information governance, requiring the Private Knowledge Safety Fee to help sectoral regulators in avoiding pointless processing of non-public information throughout AI analysis and utility and selling private information safety by default and by design measures or mechanisms. It emphasises the significance of stopping extreme use of non-public information and advocates for the precept of information minimisation, whereas additionally encouraging the openness and reuse of non-sensitive information.

In June 2025, the Ministry of Digital Affairs introduced the draft Act for the Promotion of Knowledge Innovation and Utilization and initiated a 60-day public remark interval to collect suggestions. The draft goals to ascertain a authorized framework for information entry and usefulness, create mechanisms for information openness, sharing, and reuse, and thereby foster technological innovation and R&D, together with within the subject of AI.

The AI regulatory framework in Taiwan remains to be below improvement. At the moment, neither the Draft AI Primary Act nor these non-binding AI tips impose obligatory obligations on the non-public sector or grant enforcement energy to competent authorities. Nonetheless, as soon as the AI Primary Act is enacted, sectoral regulators are anticipated to formulate sector-specific AI rules.

As of now, there are not any vital AI-related non-public litigation instances in Taiwan. Nonetheless, as AI applied sciences proceed to proliferate, disputes over AI points resembling mental property infringement and algorithmic bias are anticipated to come up.

At the moment, three main AI techniques which have been applied or are into account to be used in courts and the judiciary to boost the effectivity and transparency of the judicial course of. The AI-prediction system for judicial decision-making relating to youngster custody, developed by Nationwide Tsing-Hua College and printed in September 2019, utilises case information to supply a probability-based prediction for custody outcomes. It’s supposed to help events concerned in household instances, in addition to attorneys, social employees and mediators, in understanding the rules that judges apply in parental rights instances.

The AI sentencing data system, launched in February 2023 by the Judicial Yuan, goals to harmonise sentencing practices throughout courts by analysing earlier rulings and mechanically tagging related sentencing components. The system has each factual and evaluative modes and is out there to be used by legal professionals and the general public. For instance, a defendant used the AI Sentencing Info System to know potential sentencing throughout an enchantment, as talked about in a judgment.

The AI judgment-drafting system, pre-launched by the Judicial Yuan in late 2023, goals to help judges in drafting felony judgments for DUI and fraud instances. The generative AI-based system has confronted criticism, prompting the Judicial Yuan to make clear that it solely assists judges in producing preliminary drafts primarily based on particular authorized circumstances and proof. The final word decision-making, together with fact-finding and sentencing, stays solely below the decide’s authority. The Judicial Yuan plans to conduct exterior consultations and challenge steering on utilizing the system to make sure that it adheres to the best requirements of judicial accountability.

In recent times, the authorized business has seen a big rise in using applied sciences resembling AI, generally known as authorized tech. This pattern has led to the adoption of progressive companies resembling AI contract overview and danger evaluation throughout numerous sectors, together with finance. Nonetheless, the necessity to regulate regulatory requirements in response to authorized tech stays a vital challenge. As an illustration, in 2022, the Taiwan Bar Affiliation revised its rules to permit legal professionals to promote their companies on permitted platforms, emphasising the significance of adapting regulatory requirements to accommodate new service fashions.

As an alternative of enacting legal guidelines and rules to control using AI inside the non-public sector, at this stage, the Taiwan authorities prefers to challenge tips that function non-binding administrative steering. This method gives flexibility and encourages business sectors to have interaction in self-regulation, aiming to foster innovation and flexibility inside the quickly evolving subject of AI. The Draft AI Primary Act additionally permits authorities companies discretion and adaptability to align with worldwide requirements and practices by formulating the AI danger classification framework and associated rules sooner or later.

The Draft AI Primary Act and AI tips mirror the Taiwan authorities’s proactive and strategic efforts to deal with AI-related dangers whereas selling AI expertise innovation and improvement. By sustaining a forward-looking regulatory framework, Taiwan is properly ready to navigate the complexities of the AI period.

Timeline

• 2017 – Taiwan launches AI Grand Strategy for Small Nation undertaking to put money into creating AI ecosystem
• 2018 – Govt Yuan rolls out Taiwan AI Motion Plan 1.0 (2018-2021) to boost AI capabilities
• September 2019 – Nationwide Science and Know-how Council points AI Know-how R&D Tips for researchers’ reference
• 2022 – Taiwan Bar Affiliation revises rules to permit legal professionals to promote companies on permitted platforms
• Late 2022 – Introduction of ChatGPT 3.0 strengthens Taiwan’s function in offering {hardware} infrastructure
• February 2023 – Judicial Yuan introduces AI sentencing data system to harmonise practices throughout courts
• June 2023 – Govt Yuan approves Taiwan AI Motion Plan 2.0 (2023-2026) specializing in expertise improvement
• August 2023 – Govt Yuan points Public Sector Gen AI Tips establishing framework for moral utility
• December 2023 – Truthful Commerce Fee publishes White Paper on Competitors Coverage in Digital Economic system
• March 2024 – Ministry of Digital Affairs releases Draft AI Analysis Tips classifying AI into danger ranges
• April 2024 – Truthful Commerce Fee fines Agoda for utilizing competitor’s enterprise identify as key phrase in AI-powered promoting
• June 2024 – Monetary Supervisory Fee releases Tips for Use of AI in Monetary Trade
• July 2024 – Taiwan authorities enacts Statute of Fraud Crime Hurt Prevention mandating deepfake disclosure
• August 2025 – Executive Yuan submits Draft AI Basic Act to Legislative Yuan on August 28 designating Ministry of Digital Affairs as competent authority
• November 2025 – Report on Taiwan AI law published detailing complete framework drafted previous to Legislative Yuan overview

Abstract

Who: Taiwan’s Govt Yuan, Legislative Yuan, Ministry of Digital Affairs, Nationwide Science and Know-how Council, Monetary Supervisory Fee, Truthful Commerce Fee, and Lee and Li Attorneys at Regulation (Ken-Ying Tseng and Yi-Mei Pan authored the evaluation).

What: The Draft Synthetic Intelligence Primary Act establishes a governance framework for AI improvement in Taiwan, outlining 15 coverage goals together with useful resource allocation for AI industries, regulatory changes, worldwide cooperation, AI training promotion, danger prevention measures, and institution of danger classification frameworks aligned with worldwide requirements. The laws designates the Ministry of Digital Affairs as competent authority and requires authorities companies to conduct danger assessments when utilizing AI techniques.

When: The Govt Yuan permitted and submitted the Draft AI Primary Act to the Legislative Yuan on August 28, 2025, following years of preparatory work together with Taiwan AI Motion Plan 1.0 (2018-2021), Taiwan AI Motion Plan 2.0 (2023-2026), and a number of guideline publications all through 2023-2024. The evaluation doc was drafted in November 2025 previous to Legislative Yuan finalisation.

The place: The regulatory framework applies all through Taiwan, affecting the semiconductor business (projected manufacturing worth exceeding NT$6.4 trillion in 2025), monetary establishments, authorities companies, and AI startups. Taiwan’s strategic place within the international {hardware} provide chain for computational processing energy positions the island as a vital participant in AI infrastructure improvement.

Why: Taiwan developed this complete regulatory method to steadiness innovation with danger administration as AI functions proliferated following generative AI’s emergence in late 2022. The framework addresses issues about algorithmic bias, information governance, mental property rights, legal responsibility attribution, client safety, and competitors points whereas sustaining Taiwan’s technological competitiveness. The laws responds to international AI governance tendencies whereas leveraging Taiwan’s semiconductor dominance to advance AI ecosystem improvement by way of non-binding steering quite than prescriptive regulation.