AI Japan Launches LEAK GUARD Chrome Extension For Enterprise AI Input
The extension treats employee use of generative AI as something to be inspected at the moment of sending rather than prohibited, which places the control point at the browser instead of the network.
Reporting from 1 source: ASCII.jp.
AI Japan, based in Minato-ku, Tokyo, has begun offering LEAK GUARD, a Chrome extension for enterprises that inspects input content right before it is sent to generative AI services. API keys, credit card numbers, and My Number identifiers block the send. Email addresses and phone numbers trigger a warning. Inspection runs on-device, and original text is not sent to LEAK GUARD's servers.
The inspection runs in the browser at the instant a user presses Enter or clicks send, and it reads the whole input field once rather than monitoring continuously. A protection indicator sits at the bottom right of the ChatGPT screen while the extension is active.
Detections are handled by category. API keys and access tokens, credit card numbers, and My Number identifiers stop the original text from being sent. Email addresses and phone numbers produce a warning, and the user can confirm to send once or cancel. Attached files are not read; only file names and extensions are checked.
Only masked body text is stored when a detection occurs, with the detected value replaced by a category label such as [REDACTED:EMAIL]. Administrators see a list of time, user, information type, and whether the event was a warning or a block. Body text is shown only after an explicit administrator action, and that viewing is written to an audit log.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.