Browser spell-checkers face a sharper permission question

A browser spell-checker has to see the words it is meant to correct. That basic fact makes site permissions and data handling part of the product, not a footnote to accuracy. For builders, the important question is not simply whether a suggestion is useful. It is what text the extension can access, when it accesses it, and whether any of that text leaves the browser.

TypoGuard describes itself as a Chrome extension for correcting text in chat boxes and contact fields before submission. Its published materials say its correction technology uses Gemini AI through Google’s API. That disclosure identifies a cloud service in the correction path. It does not, on its own, explain which text is sent, how requests are scoped, or what happens to submitted text afterward. Those are separate questions, and users should not have to infer the answers from a general description of the feature.

Permissions are part of the correction model

Browser extensions need access to pages or fields to work in them. The scope of that access matters. A tool intended to help in supported writing fields has a different trust boundary from a tool that can inspect broad browsing activity. The permission prompt is an early signal, but it is not a complete account of runtime behavior or data handling.

Builders should describe permissions in terms users can connect to an action. Which supported sites or page elements can the extension read? Does it act only when a user requests a correction, or can it inspect text as it is typed? Is text handled locally, sent to an API, or both? Does the extension retain request content? If the answer varies by feature or plan, say so plainly. “Works in text fields” does not answer these questions.

TypoGuard names supported web platforms and says it corrects text before it is sent. That is useful product context, but it should not be treated as proof that the correction runs locally or that only a selected passage is processed. Its disclosure of Gemini through Google’s API is the firmer boundary to note. The public description supplied here does not specify retention, encryption, or the exact payload sent, so users should consult the product’s current privacy materials rather than assume details.

Cloud correction changes the disclosure burden

Sending text to a cloud service can support more capable suggestions, but it introduces another party into the data path. A misspelled sentence may contain customer details, a private message, or information copied into a form. The correction layer sits before submission, yet the text may still leave the browser during correction. “Before you hit send” describes timing relative to the recipient; it does not establish where processing occurs.

That distinction is especially relevant in work accounts and regulated settings. A team evaluating an extension should check its data disclosures alongside its permission request, then compare those statements with internal rules for confidential text. If policy bars sending certain content to external services, a browser-based assistant that uses a cloud API may not be suitable for those fields. A clear opt-out or a way to avoid sensitive pages can help, but builders should claim such controls only when they exist and explain how they behave.

What the public product details establish

There is no verified release or pricing change to report from the available product information. The useful update for a buyer is the currently stated scope and price, not an assumed market move. TypoGuard’s homepage says the free offering includes 100 corrections per month and requires no account. Its pricing section describes the free plan as limited, lists Pro at $6.99 billed monthly, and lists Ultimate at $49.99 billed yearly. Those are published terms, not evidence that prices changed this month.

The same materials say paid plans include unlimited corrections, additional supported languages, priority ticket support, and faster correction responses. They identify Chrome and Microsoft Edge in the product description, while the detailed product presentation focuses on Chrome. That is enough to prompt practical checks: confirm browser compatibility, supported sites, and the exact plan terms before rolling the tool out. It is not enough to assume identical coverage across every site or browser.

A short audit for builders

• Map access: Explain which pages and text fields the extension can inspect, and why those permissions are needed.
• Map processing: State when a correction request is sent to a cloud API and what text is included.
• Map retention: Give a direct answer about storage, retention, and deletion. If a detail is not established, do not imply a promise.
• Test sensitive workflows: Check behavior in customer-support, booking, and internal tools before recommending broad deployment.

Browser writing assistants compete on convenience, but convenience is not the whole evaluation. The practical standard is legibility: users should be able to understand what the extension can read and where their text goes before they install it or type sensitive material. For TypoGuard, the public product description establishes a pre-submission correction workflow and use of Google’s API. The remaining data-handling specifics are the questions a careful user should resolve before relying on it for confidential writing.