Generative AI is being explored for use across a wide range of fields, and the intellectual property (IP) sector is no exception.

Given its ability to process and generate language-based content, generative AI is considered highly compatible with tasks such as patent searching and patent analytics.

On the other hand, when it comes to drafting patent specifications, several challenges have been identified, including the following:

(1) Patent specifications require strict legal precision and accurate use of technical terminology.

(2) Confidential technical information must be carefully protected to prevent any risk of data leakage.

As a result, the practical use of generative AI in this area appears to remain limited at present.

Amid such circumstances, more than 3,500 patent applications filed by SoftBank Group were published over just two days, on April 2 and April 3, 2025—an event that drew attention within the IP community (see the link below).  According to our search of the patent database, all of these applications were filed by just two patent firms.  Handling such a large volume using conventional methods would typically be extremely difficult, which suggests that the company may be applying this technology to the preparation of patent specifications.

It may not be long before this technology significantly enhances the efficiency of patent drafting tasks in the future.

https://xtech.nikkei.com/atcl/nxt/column/18/03093/040200003/ (Japanese only)