Browser & SaaS LLM

ChatGPT, Claude, and similar web apps — session, cookie, upload, and vendor training-risk visibility

Customer POC framework

Extension preferred — or simulates workforce/browser GenAI telemetry.

Why this is different from “local model training”

When you use ChatGPT in the browser, your data is sent to OpenAI's service. The vendor may retain or use uploads under their terms — that is not the same as training a model file on your laptop (Ollama / Hugging Face). This console shows browser session signals (history, cookie store updates) and upload events so governance teams can treat it as shadow AI / data disclosure, then route users to enterprise connectors with block policies when ready.

Recommended: Download the browser agent from AI Guard — per-machine logs, visibility-first monitor mode, then Warn/Block.

Or run: .\scripts\scan-browser-llm.ps1 for history/cookie metadata only (close Chrome if history is locked).

Click sessions or telemetry events — primary shadow path for data sent to vendor LLMs.

Detected browser LLM sessions

No sessions yet — run the browser scanner after using ChatGPT in Chrome or Edge.

Telemetry (live — refreshes every 2s)

PII events (e.g. Aadhaar) appear as pii detected as soon as the extension syncs to this console.