Digital Arts adds four new features to AI Chat Filter, which shows how generative AI, such as ChatGPT, are being used and ensures safe use. — Comes standard with i-FILTER as of November 13 —

　Information security solutions provider Digital Arts Inc. (headquarters: Chiyoda-ku, Tokyo, Japan; CEO: Toshio Dogu; hereinafter referred to as “Digital Arts”; Code 2326) is pleased to announce new features coming to AI Chat Filter, which is a standard feature in the i-FILTER series that provides support for generative AI (Artificial Intelligence), on November 13, 2023.

AI Chat Filter for the Growing Use of Generative AI

　Digital Arts has developed an AI Chat Filter that allows users to control generative AI being used and visualize how they are being used. The feature became a standard feature in i-FILTER and i-FILTER@Cloud on August 16, 2023. Both products utilize a database of URLs collected over years to create a system called Whitelisting, which only allows access to websites that Digital Arts has deemed safe.

　AI Chat Filter displays a log of information, such as when, who, and what kind of prompts were sent to generative AI. There is also a filter that blocks prompts that contain certain phrases, such as confidential information. The blocked words can also be viewed from the log. When using generative AI, a warning screen also pops up to remind users to be careful not to leak confidential information. These features make it possible to check that generative AI is being used appropriately.
　It also has the ability to flexibly control the use of generative AI. Using our Web Service Controls that are a standard feature of the i-FILTER series, viewing of generative AI such as ChatGPT, Bard, and Bing AI can be allowed or blocked, Group Settings can control which groups or users can access generative AI and Time Controls can allow access to generative AI only during work or class hours.

　With this update, we will add four new features to AI Chat Filter to make the use of generative AI safer and smarter.

Four New Features for Smarter, Safer Use of Generative AI

　The four new features being added are "AI chat log (responses)," "Masking," "Word templates" and "DLP scanning."
　The AI chat log (responses) feature logs the responses received from the generative AI. The ability to log both outgoing and incoming messages in AI chat logs enables a more detailed understanding of usage and visualization of data, contributing to early detection of potential information leaks.
　The masking feature makes it possible to mask (hide) certain words that have been set in advance from prompts being sent to the AI. When keywords that may link to personal or confidential information are included, this feature allows users to replace the text instead of blocking the prompt, thereby reducing the risk of information leaks while still maintaining convenience.
　Word templates are templates of words that Digital Arts recommends users block. Using these templates removes the need for creating a list of words to block and makes installing AI Chat Filter that much easier.
　DLP scanning scans prompts sent to generative AI to block those that may contain personal or sensitive information When used in combination with masking, if the system detects that personal or confidential information is input, it can automatically mask the relevant parts of the prompt before sending. This makes it possible to prevent information leaks before they occur.

By expanding the features of AI Chat Filter, Digital Arts hopes to support the safe use of generative AI more than ever.