Product

Turrigan Guard

Turrigan Guard is a browser extension that checks your prompt on your own device the moment you press Send, and pauses it if it contains personal data. It works in ChatGPT, Claude, Gemini and Microsoft Copilot. The Turrigan platform governs what AI says and does. Guard governs what goes into it.

Turrigan Guard is on the Chrome Web Store, published by 8plus2 Omnitech OPC Private Limited. The Personal edition is free.

The problem Guard solves

Most AI data exposure is not an attack. It is an employee pasting a customer record, a card number or an ID into a chat box because it was the fastest way to get an answer. Network filters see an encrypted request to a domain the company has approved. Policy documents ask people to be careful. Neither one is in the room at the moment the message is sent.

Guard is. It reads the composer, not your traffic, and it acts at the last moment where the decision is still reversible: after you have written the message, before it leaves the browser.

How it works

The check runs entirely on your device. There is no cloud call in the path, which is also why it is fast enough to sit in front of the Send key.

Two editions

Personal, free

A separate extension that collects nothing, stores nothing and sends nothing anywhere. It has no network permission and no background component, so it cannot transmit data even in principle. There is no analytics, no tracking, no account and no server. We receive no information about you or your use of it.

Enterprise, paid

The same on-device block, plus an audit trail your organization owns. When a user proceeds past both review prompts, the extension records one minimised event in your Turrigan tenant. The event contains the detector types and counts, a masked sample such as email · ••••.com, the app host, a timestamp and the extension version. It never contains the prompt text and never contains a raw detected value. Turrigan rejects any event whose sample is not masked, so a misbuilt client cannot land raw personal data in the log.

Each event is appended to your tenant's hash-chained guard-events log, the same integrity backbone as the decision log, so the audit is independently verifiable. Events are isolated per tenant, and a re-sent event is de-duplicated on its client event id, so a network blip never double-records.

Deploying Enterprise

Three things, all set centrally through Chrome or Edge managed policy so IT owns them and users cannot change them: the enterprise mode flag, your Turrigan ingest endpoint, and a scoped guard:ingest API key that can post guard events and read its own entitlement and nothing else. It cannot read decisions and it cannot govern.

If the subscription lapses, the extension degrades to on-device blocking only. Users stay protected; audit logging and central management stop until you renew. An unreachable entitlement check is tolerated for a seven-day offline grace window, so a transient outage does not disable a paying tenant. It does not go dark and leave people unprotected.

Frequently asked questions

Does Turrigan Guard send my prompts anywhere?

No. The Personal edition has no network permission at all. The Enterprise edition never sends prompt text either. It sends a minimised, masked event to your own organization's Turrigan tenant, not to us.

Which AI apps does it support?

ChatGPT, Claude, Gemini and Microsoft Copilot, in Chrome and Edge.

How is Guard different from the Turrigan platform?

The platform governs AI outputs and agent actions through the /v1/govern endpoint. Guard governs AI inputs. The platform is what your systems call. Guard is what your people use. See about Turrigan for how the products fit together.

Where is the privacy policy?

Both editions are covered by the Turrigan Guard privacy policy.

Install Turrigan Guard for Chrome Ask about Enterprise