Changes
Most recent changes first. Version 0.1.0 comes from the Chrome Web Store; the browser installs new versions automatically.
On this page
18 October 2026
- Install from the Chrome Web Store for Chrome and Edge: one click on "Add to Chrome", no ZIP file and no developer mode (Chrome, Edge).
- Companies force-install via
ExtensionInstallForcelistwith the extension IDlefajbhockjgdhaohnacgaggfepianjh(For companies). - Open source under the GNU AGPL v3: the source code is on GitHub.
- New name: AImori, with the kabuto helmet as its logo.
New in the store version:
- 242 data types in 60 packs instead of 57 in 13 (What gets detected).
- Image checks: AImori reads text in pasted or uploaded screenshots locally with text recognition and checks it; if something is found, a redacted copy can go to the site instead.
- Local AI detection: a model on your device also detects names, companies, addresses and health information. The rules always apply.
- Placeholders instead of redacting, e.g.
[IBAN_1]. The AI's answer can then be shown with the real values – only in a view of the extension, never on the AI site. - Rephrase safely: replaces sensitive details with general descriptions; the question stays the same.
- Not sensitive: values marked this way are not reported again for 30 days, on any site.
- Slim bar: Redact and Edit stay visible; Placeholder, Rephrase safely, Not sensitive and Send anyway are under "⋯ More options".
- AI training indicator: a coloured dot on the icon shows whether an AI site uses inputs for training by default. In the popup: "Check AI training of a site".
- Voice mode: spoken conversations with the AI cannot be checked; AImori warns before it starts, companies can block it.
- Hidden instructions: a warning when pasted text or uploaded text documents show signs of them.
- Confidential sources: text copied on specified sites (e.g. banking, HR) is stopped when pasted into an AI.
- Form guard: forms on specified sites are checked before they are submitted.
- For companies by policy: control AI sites (notice or block on AI sites that are not approved, with a button to the approved AI) and rules per account (separate rules for the business and personal account of the same AI), see For companies.
14 October 2026
- Messages as a narrow bar above the input field instead of a large dialog: "Sending paused: IBAN detected" with Redact, Edit and Send anyway (two clicks instead of a checkbox). On paste: Insert redacted, Discard, Insert anyway (Warning bar and actions).
- Popup and settings in a new design; settings save immediately.
- Automatic country packs: AImori chooses the country packs itself from time zone and language region – locally, without network, without location. Override in the settings with "Choose myself" (Country packs).
- New policy
autoCountryPacks: locks the mode to "Automatic" or "Choose myself".
13 October 2026
- Warning bar while typing: after a short pause in typing, a narrow bar above the input field shows what was detected, with "Redact now" (Warning bar and actions).
- Microsoft 365 Copilot added: copilot.cloud.microsoft and the Copilot chat field on m365.cloud.microsoft.
- Send blocking on claude.ai and current ChatGPT detected more reliably.
Version 0.1.0 (13 October 2026)
First release:
- Detection of 57 data types in 13 packs, with check digits (What gets detected).
- Checks on send and paste with redact, cancel (today: edit) and send anyway (with confirmation).
- Popup with counters, settings page, six languages.
- Policies, branding (white label), ADMX templates for Edge and Chrome, PowerShell scripts for Intune and registry (For companies).