Guardia Companion

Guardia Companion — Browser Extension Description Guardia Companion is the browser-based companion to Guardia, a child-safety platform that helps parents understand what their kids are actually doing when they talk to AI chatbots on the web. Once a parent pairs this extension with their child's browser using a one-time code from their Guardia TrustEd dashboard, the extension quietly watches the child's conversations on supported AI chat websites — currently Google Gemini, ChatGPT, and Claude — and reports the real, turn-by-turn conversation back to the parent's dashboard, where Guardia's GuardRail engine scans every message for safety concerns before the parent ever sees it. WHY THIS EXISTS AI chatbots have become one of the most common places kids spend time online, whether for homework help, casual conversation, or creative writing. Unlike a search engine, an AI chatbot can hold an extended, private, back-and-forth conversation with a child, and that conversation can drift into territory a parent would want to know about — anything from a child asking a chatbot to help them get around a school's content filter, to a chatbot inadvertently walking a child through something age-inappropriate, to more serious signals like self-harm language or bullying. Most parental control tools block access to sites entirely or filter by keyword at the network level, which either stops a child from using a legitimately useful educational tool at all, or misses context that only shows up inside an actual conversation. Guardia takes a different approach: let the child use the AI tools their school or family has approved, but make sure a parent can see the real transcript, with real safety scoring, without having to read over their child's shoulder all day. WHAT IT ACTUALLY DOES Once paired, the extension runs quietly in the background on the specific AI chat sites it supports. When your child sends a message and the AI replies, the extension reads the text that is already visible on the screen — the same words the child typed and the same words the AI replied with — and sends that text, along with which site it came from, to your family's private Guardia account. It does not read anything on any other website. It does not read email, social media, banking sites, or anything outside of the specific AI chat domains it is built for. It does not take screenshots, record video, or log keystrokes as they are typed; it only reads the finished message once it has actually been sent or received, the same way a parent glancing at the screen would see it. Every message that reaches your Guardia account is scanned automatically by GuardRail, Guardia's compliance and safety-scoring engine. Messages are checked against grade-appropriate policies and assigned a risk level. If something concerning comes up, a flag appears in your TrustEd parent dashboard along with the full surrounding context, not just an isolated snippet — because a single line pulled out of context can be misleading, and a parent deserves to see the real conversation to make an informed judgment. Routine, unremarkable conversations — the vast majority of them — are also viewable if you want to see them, since transparency is the whole point, but nothing about the design tries to alarm you over ordinary chatbot use. WHO CAN SEE THE DATA Only the parent or guardian account that generated the pairing code can see what this extension reports. Data goes directly from the extension to your family's Guardia account over an encrypted connection; it is not sold, shared with advertisers, or used to train any AI model. The extension itself stores only a small pairing token locally in the browser, used to authenticate its reports to your account — it does not store a copy of the conversations locally beyond what is needed to send them. HOW PAIRING WORKS A parent generates a short pairing code from the TrustEd dashboard, tied to a specific child profile. That code is entered once into this extension's popup, on the browser the child actually uses. From that point forward, the extension is linked to that child's profile specifically, and everything it reports is filed under that child's account, alongside anything reported by other Guardia-connected devices, such as the Android companion app, so a parent gets one unified view instead of several disconnected ones. UNPAIRING AND HONEST LIMITATIONS A parent can set an unpair password from the TrustEd dashboard. If set, that password must be entered in the extension's own popup before it will disconnect. We want to be completely upfront about what this does and does not do: this only gates the extension's own unpair button. Chrome, by design, allows any extension to be removed or disabled entirely from the chrome://extensions page, and no browser extension — including this one — can prevent that. If a child has full administrative access to the computer they are using, they can remove this extension the same way they could remove any other program. This extension is a transparency tool for families who want visibility, not a tamper-proof enforcement mechanism, and we think it is more honest to say that plainly than to imply a level of lockdown that Chrome's architecture does not actually allow any extension to provide. For situations where stronger enforcement matters, we recommend pairing this with your organization's or household's device management policies, such as a Google Workspace for Education managed Chrome profile, which can enforce extensions at the operating-system level in a way no individual extension is able to do on its own. WHAT SITES ARE SUPPORTED At present, the extension actively monitors conversations on gemini.google.com, chatgpt.com (and the legacy chat.openai.com domain), and claude.ai. These were chosen because they are, by a wide margin, the most commonly used general-purpose AI chat assistants among school-age children today. We are actively working on expanding coverage as usag…