Does PrivacyScrubber send my data anywhere?

No. All detection and masking happens locally in your browser's memory. You can verify this yourself by opening your browser's Network tab or disconnecting from the internet — the tool keeps working.

What AI tools does it work with?

ChatGPT, Claude, Gemini, Copilot, Grok, Llama, Perplexity, Mistral, and Hugging Face Chat. The browser extension adds in-page masking directly inside each of these.

Is the free version good enough for regular use?

The free tier covers core PII types (names, emails, phone numbers) up to 15,000 characters per scrub, with a 3-page PDF limit. Batch processing, OCR, and industry-specific detection profiles (legal, medical, finance, etc.) require PRO.

Does this satisfy HIPAA, GDPR, or SOC 2 requirements?

PrivacyScrubber is designed to support de-identification under HIPAA Safe Harbor, GDPR Article 25/32, and SOC 2 CC6-series controls by removing identifiers before data reaches a third-party AI tool. It is a technical control, not a substitute for a full compliance program — check with your compliance team on whether it fully satisfies your specific obligations.

Can I get my original data back after the AI responds?

Yes. Paste the AI's response back into the tool and click "Reveal" — it swaps the tokens back to your real data locally, with no server round-trip.

Is PrivacyScrubber a local or cloud PII scrubber?

PrivacyScrubber is a 100% local, browser-based PII scrubber. Unlike cloud DLP solutions, it performs enterprise-grade PII and PHI scrubbing entirely in your browser's RAM, ensuring zero-trust data sanitization for texts and documents (PDFs, DOCX, CSV) without ever sending data to an external server.

What happens to the session map when I close the browser?

The session map is stored exclusively in volatile browser RAM. The moment you close the tab, reload the page, or navigate away, the entire mapping — including all original PII values — is permanently destroyed.

Does PrivacyScrubber remove PII from uploaded documents, and what is the output format?

Yes. PrivacyScrubber detects and scrubs PII from local documents (.txt, .docx, .pdf, .xlsx, .csv, and images). You can download the output either as a fully sanitized, anonymized document (retaining its original file extension and structure) or copy the processed text stream directly to your clipboard.

How does the TEAMS encrypted session sharing work?

PrivacyScrubber generates an AES-GCM encrypted link locally. Your colleague can paste the AI's response and use your shared, encrypted key to safely 'reveal' the original data on their machine — no database or server involved.

How does closed Shadow DOM sandboxing protect my data from malicious scripts?

PrivacyScrubber encloses both your editor input (#inputText) and the highlighted token output within isolated, closed Shadow DOM roots. This makes the raw, unredacted text completely inaccessible to any third-party scripts, browser extensions, or key-sniffing scripts running in your main document pool, creating a hardened frontend sandbox.

Can corporate accounts or accounting firms pay via manual invoice or wire transfer (SWIFT/SEPA)?

Yes. If your corporate credit card is declined due to international banking restrictions or if your procurement department requires manual invoicing, purchase orders (PO), or bank wire transfers (SWIFT/SEPA), our team can manually issue a formal invoice with tax reverse-charge support. Contact support@privacyscrubber.com to request an enterprise invoice.

How does PrivacyScrubber prevent data residue in system RAM (V8 memory wiping)?

To prevent PII or case documents from remaining in your device's memory after sanitization, the processing engine actively overwrites all temporary parsing buffers (used for DOCX, PDF, and spreadsheet extraction) at the byte level using Uint8Array.fill(0) in local memory immediately after use, ensuring zero sensitive residue remains in RAM before garbage collection.

Does changing the detection profile reset my protected session data?

No. Changing your detection profile (e.g., from General PII to Wealth Management) does not clear your active session or reset your local token maps. This ensures that any tokens generated using your previous profile remain fully decryptable when you paste AI responses back to 'Reveal' the original text. To completely wipe your session, rotate encryption keys, and clear all memory, click the 'Clear' button at any time.

How do local Audit Receipts and Teams Handoff work without a backend database?

PrivacyScrubber operates entirely in your browser's local RAM. Compliance audit receipts and team handoffs are exclusive features of the Teams Subscription. For audit compliance, the system builds an offline cryptographic PDF receipt documenting entity counts, timestamps, and active compliance profiles. For Teams Handoff, it encrypts the tab's session map in-memory using Argon2id and XChaCha20-Poly1305. Your colleague decrypts the token map on their machine using the shared key — no central database or server ever touches your data.

Does PrivacyScrubber use third-party tracking, analytics, or review widgets?

No. To maintain a strict zero-trust data sanitization architecture, PrivacyScrubber does not load third-party analytics trackers, cookies, or reviews integrations (such as Google Customer Reviews or similar widgets). Adding dynamic third-party scripts would require relaxing our Content Security Policy (CSP) and could allow external servers to inspect page states or log user environments. We display reviews statically or link to external platforms like G2 or the Chrome Web Store to ensure zero data leakage.

Can I use PrivacyScrubber as a PII MCP Server for Cursor or Claude Desktop?

Yes. PrivacyScrubber offers an official local Model Context Protocol (MCP) server package (@privacyscrubber/mcp-server) that integrates directly with Cursor IDE, Windsurf, and Claude Desktop. It acts as a local prompt security gateway, scanning and masking PII and API keys locally in volatile RAM before they are transmitted to external LLM APIs.

How are Team Administrator roles managed without a backend server?

In our Zero-Trust Data Sanitization (ZTDS) architecture, there are no server-side accounts or traditional 'Admin' vs 'User' roles. Team Administration is handled via client-side Blueprints. A team lead sets up the required compliance profiles, custom regex rules, and organization name locally, then locks the configuration and exports a cryptographic 'Team Blueprint' URL. When workers click this URL, their local PrivacyScrubber instance activates the TEAMS license, imports the mandated configuration, and disables their ability to edit or delete rules, ensuring organization-wide compliance securely without a backend database.

What is the difference between Token Labels and Custom Regex Rules?

Custom Regex Rules and Token Labels serve distinct purposes in Zero-Trust AI sanitization. Custom Regex Rules are a pattern discovery engine used to identify new or proprietary identifiers (such as internal ticket numbers, project codes like PROJ-\d{4}, or custom account formats) that are not covered by default NER models. In contrast, Token Labels do not search for new text; instead, they rename the replacement placeholders for standard detected entities (for example, mapping NAME → PATIENT or ID → CASE_NUMBER) so that the LLM receives domain-accurate semantic context in its prompt without requiring custom regex code.

How do Custom Token Labels optimize LLM prompt comprehension?

When prompts are sanitized using generic placeholders like [NAME_1] or [ID_1], LLMs may lose contextual role differentiation in multi-party prompts. Custom Token Labels allow PRO and TEAMS users to alias standard entity categories to domain-specific terminology (such as [PATIENT_1], [CLIENT_1], [PLAINTIFF_1], or [DEFENDANT_1]). This gives the LLM precise semantic clarity about each entity's functional role while preserving 100% anonymization and enabling seamless 1-click reverse-scrubbing ('Reveal') of the AI response.

AI Shadow Leaks & News

Anthropic MCP Design Vulnerability: Systemic RCE & AI Supply Chain Threat

Critical architectural flaws in Anthropic's Model Context Protocol (MCP) enable remote code execution and systemic data exposure across enterprise AI integrations, halted only by local client-side sanitization.

Ilya SibiryakovPrivacy Architect

Published: August 16, 2026 · Updated: August 16, 2026 · 3 min read

100% Local Processing ✈ Airplane Mode Verified⊘ No Server Logs

Live Simulation

Zero-Trust Data Sanitization

Watch PrivacyScrubber's local engine transform sensitive Security data instantly in your browser, without any API calls.

Automated Detection Classes:

User / Server IP AddressesAWS_KEYINTERNAL_HOSTNAMEMAC_ADDRESSVULN_ID

100% Client-Side Execution

Wasm_Engine

SIEM ALERT > Src IP: 192.168.12.44 → Dst: siem.internal.corp User: d.novak@corp.com | AWS Key: AKIA4X9M2PLRT887NNZZ CVE: CVE-2026-44821 | Severity: CRITICAL

SIEM ALERT > Src IP: [IP_1] → Dst: [HOSTNAME_1] User: [EMAIL_1] | AWS Key: [API_KEY_1] CVE: [CVE_1] | Severity: CRITICAL

AI Risk Calculator

Company

AI Users50

Risk● Critical

Leaks/yr

9,000

Max Fine

€20M

Technical Incident Analysis

Recent security disclosures have revealed a systemic design flaw in Anthropic's Model Context Protocol (MCP), exposing thousands of enterprise applications to remote code execution (RCE) and automated prompt injection attacks. Because MCP servers automatically execute contextual tool calls, attackers can weaponize inputs to exfiltrate database connections, source code, and internal telemetry. Similar to previous OpenAI Hardware Security Risks, unchecked telemetry channels present critical exposure. Explore our AI Security Architecture to understand how modern protocols are being hardened.

Enterprise Blast Radius & Compliance Risks

When autonomous agents leverage compromised MCP servers, regulatory perimeters under GDPR & CCPA Compliance are immediately breached. Exposing Personally Identifiable Information (PII) and corporate secrets to third-party endpoints through automated tool chaining creates severe compliance penalties and litigation exposure.

Client-Side Mitigation via Zero-Trust Data Sanitization

Organizations can completely neutralize this supply chain risk by adopting browser-based PII filtering. By executing tokenization in WASM browser RAM with zero network latency prior to prompt transmission, sensitive payloads are replaced with secure placeholders before ever reaching vulnerable model contexts. Learn more about our Model Context Protocol (MCP) Sanitizer and sessionMap ephemeral isolation protocols.

ChatGPT & Enterprise LLMs Integration

How to Protect Data for Anthropic MCP Design Vulnerability

PrivacyScrubber operates entirely client-side. Whether using the copy-paste dashboard or the browser extension, your sensitive records stay on your local device. Follow these instructions to safely use ChatGPT & Enterprise LLMs:

1 Method A: Zero-Trust Web Workspace (Copy-Paste)

Best for manual prompt sanitization without installing plugins:

Open the PrivacyScrubber Web App dashboard in your browser.
Paste the raw prompt or document text containing sensitive customer, employee, or proprietary identifiers.
Click Protect PII. Sensitive data is instantly swapped for secure placeholders (e.g., [NAME_1]).
Submit the sanitized prompt to ChatGPT & Enterprise LLMs.
Paste the AI's answer into the Reveal Originals box to instantly restore the original values.

2 Method B: Chrome Extension (In-Context Redaction)

For automated, inline de-identification within chat interfaces:

Install the free PrivacyScrubber Chrome Extension from the Web Store.
Navigate to your AI chat interface. A PrivacyScrubber shield button will appear inline.
Paste your raw prompt. Click the shield button to sanitize all identifiers instantly in-place.
Send the prompt to the AI chatbot.
The extension automatically intercepts and detokenizes the response, displaying raw values to you.

Local Redaction & Risk Matrix for Security

Detection Entity	Token Placeholder	Risk Level	Security Action
User / Server IP Addresses	[IP_ADDRESS]	High (DLP / Location footprinting)	IPv4 / IPv6 format strip
AWS_KEY Details	[AWS_KEY]	Medium (PII Exposure)	Deterministic local swap
INTERNAL_HOSTNAME Details	[INTERNAL_HOSTNAME]	Medium (PII Exposure)	Deterministic local swap
MAC_ADDRESS Details	[MAC_ADDRESS]	Medium (PII Exposure)	Deterministic local swap
VULN_ID Details	[VULN_ID]	Medium (PII Exposure)	Deterministic local swap

VERIFIABLE WORKFLOW

From Raw Security Data to Clean AI Prompt

3 Steps, 30 Seconds, Zero Server Hops.

Open PrivacyScrubber or the Chrome Extension. Paste your raw prompt or document text. What reaches ChatGPT looks like this: [NAME_1][EMAIL_1]. Your original data stays local the entire time.

Paste Your Real Data

Paste your actual prompt or document text into PrivacyScrubber — or click the shield icon directly inside ChatGPT, Claude, or Gemini. No copy-paste workaround. No second tab. It sits right where you already work.

Automated Detection Classes:

[IP_ADDRESS][AWS_KEY][INTERNAL_HOSTNAME][MAC_ADDRESS][VULN_ID]

Names Out, Tokens In — Locally

The engine runs inside your browser. Every real name, ID, and email is replaced with a safe token ([NAME_1], [EMAIL_1]) before the prompt is sent. The AI analyzes your actual business logic — but sees zero real identities.

Safety standard:

Airplane Mode Verified (RAM Only)

Get the AI's Answer Back in Plain Language

Paste the AI's response into Reveal Originals. PrivacyScrubber swaps every token back to the original value — instantly, inside browser RAM. Close the tab and every mapping is gone. Nothing stored, nothing logged, nothing sent.

Privacy Guarantee:

Mapping destroyed on tab close

Enterprise Adoption Use Cases

CISO Security TeamDLP GOVERNANCE

Zero-Trust Verified

Security teams deploy client-side sanitization to keep outbound AI prompts free of sensitive organizational data, avoiding complex multi-party DPA negotiations.

VP of EngineeringENGINEERING SEC

Zero-Trust Verified

Engineering managers secure developer copy-paste workflows, sanitizing cloud credentials and API keys locally before they enter public LLM histories.

Risk & Audit LeadCOMPLIANCE AUDIT

Zero-Trust Verified

Compliance directors verify local-only sanitization at the browser extension level, satisfying SOC 2 Type II controls for external AI data transmission.

Data Protection OfficerGDPR COMPLIANCE

Zero-Trust Verified

Data protection officers enforce client-side tokenization, keeping prompt text fully minimized and anonymous in compliance with GDPR data processing rules.

Flat Rate — Unlimited Seats

Your Whole Team on Real Client Data. Safely. $99/mo Flat.

No per-seat pricing. No DPA negotiation. No IT portal. Secure your entire organization with client-side PII masking — $99/month flat, unlimited users. SOC 2 & HIPAA ready. Works in Airplane Mode.

Protect My Team

Zero-Trust Data Sanitization (ZTDS) — Verified Architecture

Independently auditable facts for Sensitive Data compliance teams

Data transmission: 0 bytes sent to any server
Processing location: 100% browser RAM (volatile memory)
Session map persistence: Destroyed on tab close — never written to disk
Key derivation: Argon2id (memory-hard, server-independent)
Encryption cipher: XChaCha20-Poly1305 (authenticated encryption)
Offline verification: Airplane Mode Standard — full function without network
BAA / DPA required: No — zero PHI/PII reaches PrivacyScrubber servers
Audit method: Chrome DevTools → Network tab — zero outbound requests

How to audit: Open PrivacyScrubber, enable Airplane Mode, paste any sensitive data text, click Protect PII. Open Chrome DevTools → Network tab. Zero outbound requests will confirm 100% local execution. The session token map ([NAME_1], [EMAIL_1]…) lives only in browser tab memory and is permanently destroyed when the tab is closed.

COMPLIANCE FAQ

Frequently Asked Questions

Common questions about deploying zero-trust AI for AI Shadow Leaks & News Teams.

How does this MCP design vulnerability expose enterprise data?

The vulnerability exploits auto-execution behaviors within the Model Context Protocol (MCP), allowing malicious actors to chain prompt injections into remote code execution (RCE) and siphon sensitive backend system records directly through connected AI agents.

How does PrivacyScrubber prevent this exploit?

PrivacyScrubber deploys zero-trust, client-side RAM tokenization and MCP payload scrubbing directly in the browser. By masking PII and neutralizing malicious execution arguments before requests touch upstream LLMs or agent networks, it completely halts data exfiltration vectors.

← More AI Shadow Leaks & News Solutions

Add to Device Home Screen

Anthropic MCP Design Vulnerability: Systemic RCE & AI Supply Chain Threat

Zero-Trust Data Sanitization

Try PII Masking Live

AI Risk Calculator

Technical Incident Analysis

Enterprise Blast Radius & Compliance Risks

Client-Side Mitigation via Zero-Trust Data Sanitization

How to Protect Data for Anthropic MCP Design Vulnerability

1 Method A: Zero-Trust Web Workspace (Copy-Paste)

2 Method B: Chrome Extension (In-Context Redaction)

Local Redaction & Risk Matrix for Security

From Raw Security Data to Clean AI Prompt

Paste Your Real Data

Names Out, Tokens In — Locally

Get the AI's Answer Back in Plain Language

Enterprise Adoption Use Cases

Your Whole Team on Real Client Data. Safely. $99/mo Flat.

Zero-Trust Data Sanitization (ZTDS) — Verified Architecture

Frequently Asked Questions

Save Output Package

Blueprint Generated

Add to Device Home Screen

Add to Device Home Screen

Anthropic MCP Design Vulnerability: Systemic RCE & AI Supply Chain Threat

Zero-Trust Data Sanitization

Try PII Masking Live

AI Risk Calculator

Get Your Risk Estimate

Report Generated!

Automate Compliance For Your Team

Technical Incident Analysis

Enterprise Blast Radius & Compliance Risks

Client-Side Mitigation via Zero-Trust Data Sanitization

How to Protect Data for Anthropic MCP Design Vulnerability

1 Method A: Zero-Trust Web Workspace (Copy-Paste)

2 Method B: Chrome Extension (In-Context Redaction)

Local Redaction & Risk Matrix for Security

From Raw Security Data to Clean AI Prompt

Paste Your Real Data

Names Out, Tokens In — Locally

Get the AI's Answer Back in Plain Language

Enterprise Adoption Use Cases

Your Whole Team on Real Client Data. Safely. $99/mo Flat.

Zero-Trust Data Sanitization (ZTDS) — Verified Architecture

Frequently Asked Questions

Session Memory Handoff & Governance

Save Output Package

Blueprint Generated

Add to Device Home Screen