Tidewall

Stops API keys and personal data going into ChatGPT, Copilot and Claude

Visit website

FreeAISecurityBrowser Extension11 impressionslaunches Oct 201 comment

>log in to comment

Tejveer Singh[maker]· 1d ago
Hi DevHunt 👋 I built Tidewall after seeing how easily real customer details and API keys end up in AI chats. Usually nobody means to: it’s a quick paste to get help with an email or a bug. I wanted something that stays out of the way: it runs entirely in your browser, takes about a millisecond per message, and only shows up when it finds something. I’d love your feedback on: – false positives (anything it flags that it shouldn’t) – which AI sites or key types to add next Free for individuals: Chrome and Edge, links on the site. Happy to answer anything here.

AI chatbots are useful, until someone pastes a live API key, a customer’s email or a patient’s NHS number into one. Tidewall checks each message when you press Send, on 20+ AI tools: ChatGPT, Microsoft Copilot, Gemini, Claude, Perplexity, DeepSeek, Mistral, Grok, v0, Bolt, Lovable and more. If it finds something sensitive, it holds the message and shows you exactly what the AI would see. One click swaps names, emails and numbers for placeholders like [EMAIL_1] and strips secrets out completely. When the reply comes back, the real details are put back on your screen only. The AI never sees them. What it catches: – Secrets: AWS, GitHub, GitLab, Stripe, OpenAI, Anthropic, Google, Slack and Azure keys, JWTs and private keys – Personal data: emails, phone numbers, NHS and National Insurance numbers, payment cards and IBANs (checked with real checksums, so fewer false alarms) – Your own client or project codes How it works: a Manifest V3 extension that checks everything locally, in about a millisecond. No servers, no analytics, no network calls. Its only permission is storage, plus access to the AI sites it protects. What it doesn’t cover (yet): desktop AI apps, IDE assistants and phone browsers. Free for individuals. The Team plan adds central rules (warn, redact or block for each type of data, plus approved and blocked AI tools), rolled out with Intune or Google Workspace so staff can’t switch it off.

AI Security Browser Extension

Checks messages typed into AI chat tools for personal data
Allows redaction, editing or blocking of sensitive content
Runs entirely inside your browser for privacy
Available as a Chrome and Edge extension
Reduces risk of sharing sensitive data with AI tools

Tidewall redacts personal data and secrets from messages before they reach AI chat tools, running entirely in the browser.

for: Individuals, teams and enterprises that use AI chatbots and need to protect sensitive data.
pricing: freemium

works withChromeEdgeGoogle WorkspaceMicrosoft IntuneChatGPTClaudeMicrosoft CopilotGeminiPerplexityDeepSeekMistralGrokMeta AINotebookLM

Key features

6 features of Tidewall

In-browser detection — All checks run locally in the browser, no data is sent to Tidewall servers.
Real-time redaction — When sensitive data is found, it is replaced with placeholders before sending.
Checksum validation — Numbers like NHS, payment cards and IBANs are verified with real checksums to avoid false alarms.
Custom rules & patterns — Admins can define custom client codes, project names and set allow/warn/redact/block actions.
Approved/blocked AI tools — Mark specific AI services as allowed or blocked and enforce policies per tool.
Team rollout via Google Workspace or Intune — Deploy the extension and policies to all browsers in a organization quickly.

Use cases

Prevent accidental leakage of patient NHS numbers when drafting emails with ChatGPT.
Redact API keys before sending code snippets to Claude for debugging.
Enforce company policy by blocking unapproved AI tools on corporate browsers.
Automatically replace personal email addresses in support tickets before AI summarisation.