To encourage enterprises to enhance R&D and application of AI technologies and to strengthen AI patent portfolio strategies, TIPO conducted an analysis of recent trends in global AI patent applications. The analysis covers overall AI technologies and eight specific subcategories: AI hardware, knowledge processing, machine learning, evolutionary computation, vision, natural language processing, speech, and planning and control. Utilizing TIPO’s Global Patent Search System (GPSS), the findings aim to help enterprises better understand potential development opportunities and challenges in the AI field.

The global analysis reveals that the number of AI-related patent applications has grown significantly over the past decade, surpassing 30,000 applications in 2020. Among the subcategories, machine learning and machine vision are the areas with the highest number of patent applications across countries.

In Taiwan, filing trends mirror global patterns, with applications increasing significantly since 2017. Compared to 2014, the number of AI-related patent applications in 2021 has grown nearly ninefold, with machine learning and machine vision also being the most commonly filed categories.

Taiwan ranks 9th globally in terms of total AI-related patent filings. Key Taiwanese applicants include Hon Hai (ranked 1st), ITRI (2nd), Chunghwa Telecom (4th), Institute For Information Industry (III) (7th), and Inventec (9th).

To clarify TIPO’s practical approach in examining AI-related inventions—particularly with respect to patent eligibility, enablement, and inventive step—TIPO compiled six representative case studies into the publication titled Compilation of Taiwan's AI-Related Invention Cases. On June 20, 2025, TIPO invited experts from industry, academia, and research sectors to discuss the cases. Based on their feedback, the content was refined and later introduced to the industry at a promotion and briefing session held on September 18, 2025. The publication was formally released on October 13, 2025.

This compilation is expected to help industries better understand examination practices for AI-related inventions, improve the quality of patent specifications, and strengthen global AI patent portfolios.