AgentCloak launches free AgentCloak Desktop to cloak data before AI chatbots
AgentCloak announced a free product called AgentCloak Desktop that replaces private fields in user prompts with synthetic digital twins on the user’s device before the text reaches popular AI chatbots such as ChatGPT, Claude, Gemini, Grok, Copilot and DeepSeek. The tool restores the original values in the model’s reply, so the user sees a normal answer while the real data never leaves the device.
What you can do with it
AI at Work →Replaces private data with synthetic twins on-device before sending prompts to AI chatbots and restores original values in responses.
- Who for
- founders, operations and support teams
- Cost
- free
- Effort
- minutes
Use it for
- Protect personal data in AI chatbot prompts
- Prevent accidental data leakage to AI services
- Maintain workflow with existing AI tools
Watch out Only works on supported browsers and desktop apps; synthetic twins may not cover all data types.
Key points from the news
- AgentCloak Desktop replaces private fields with synthetic twins on-device before sending prompts to AI chatbots.
- Supports ChatGPT, Claude, Gemini, Grok, Copilot, DeepSeek and runs as a browser extension or native app.
- Free for individuals; enterprise AgentCloak Server adds file cloaking, audit logs, and integrations with Microsoft Entra, Google Workspace, Okta.
The desktop version runs as a browser extension for Chrome, Safari and Edge or as a native app for macOS and Windows, and it supports English, Spanish, French, German, Italian, Portuguese and Dutch. It uses a compact on‑device model called Rampart together with deterministic rules to detect names, addresses, email addresses, phone numbers, URLs, tax IDs, bank account numbers and other sensitive fields. AgentCloak also offers a paid Server version for teams that adds file cloaking, audit logs and integrations with Microsoft Entra, Google Workspace and Okta.
AgentCloak Launches AgentCloak Desktop, Free Sovereign AI That Cloaks Private Data Before It Reaches Popular AI Chatbots
aol.com · 18 September 2026
SAN FRANCISCO--(BUSINESS WIRE)--Sep 18, 2026--
AgentCloak, a CB100 2026 top AI company, today announced AgentCloak Desktop, a free Sovereign AI product that turns the chatbots people already use into privacy-preserving AI. AgentCloak Desktop replaces private data with synthetic substitutes on the user’s device before anything goes to a frontier AI, and then restores the original values in the model’s reply.
This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20260918884045/en/
AgentCloak Desktop detects sensitive fields locally and swaps them with consistent digital twins.
As employees and individuals paste emails, contracts, customer records, and personal details into ChatGPT, Claude, Gemini, Grok, Copilot, DeepSeek, and other AI Chatbots, those prompts routinely expose names, addresses, account numbers, and identifiers to the models. Even retention promises and enterprise agreements expose sensitive data. AgentCloak Desktop addresses the problem at the source: Zero Data Sending. The model never gets the real values.
AgentCloak Desktop provides the privacy benefits of Sovereign AI while still using frontier models, without the expense and hassle of the expensive hardware, staff, and security required to maintain Sovereign AI infrastructure.
“People should not have to choose between useful AI and keeping their private data private,” said Peter Yared, Founder & CEO of AgentCloak. “AgentCloak Desktop puts Sovereign AI on the device you already have. ChatGPT, Claude, Gemini, Grok, Copilot, and Deepseek still do the heavy lifting. They just never see private data like names, addresses, or account numbers. That is Zero Data Sending, not an ambiguous policy footnote.”
Cloaked on the way out, restored on the way in
AgentCloak Desktop detects sensitive fields locally and swaps them with consistent digital twins. “Peter Parker” becomes “Julio Schmidt.” “364 23rd St” becomes “401 Vanessa Curve.” The AI still has the context it needs to answer. When the reply comes back, AgentCloak uses the same personal digital twin to put the real values back in front of the user. The original never left the device.
The product pairs deterministic rules with a compact neural model that runs entirely on the user’s computer or inside the browser. Users keep their existing AI workflows. No new chatbot. No data uploaded to AgentCloak to use the free Desktop product.
Where it runs
AgentCloak Desktop installs as a browser extension for Chrome, Safari, and Edge, or as a native app for Mac and Windows for users who use desktop AI Chatbot clients. AgentCloak Desktop adds data protection to ChatGPT, Claude, Gemini, Grok, Copilot, and DeepSeek.
What the free product detects and cloaks
First and last names Email addresses Phone numbers URLs Social Security and tax ID numbers Bank routing and account numbers Physical addresses
Languages supported in the free Desktop release: English, Spanish, French, German, Italian, Portuguese, and Dutch.
Key capabilities
Zero Data Sending: private fields are replaced on-device before the prompt is submitted; this is not a retention or “we won’t train on your data” promise.
Digital twins: synthetic substitutes stay consistent so the model can reason about people, places, and accounts without seeing the originals.
Round-trip restoration: real values are swapped back into the answer so the user sees a usable reply.
Works with the AIs people already: installs as a browser extension or desktop app rather than forcing a new, closed assistant that is expensive to run, hard to upgrade, and doesn’t connect scalably to the web.
On-device Sovereign AI: Rampart, a compact model from the U.S. government that mixes deterministic detection and neural detection of sensitive data runs completely on-device and is combined with AgentCloak digital twin technology.
Free to install for individuals; AgentCloak Server extends the same approach to teams and regulated enterprises.
AgentCloak Server adds even more power for organizations
Organizations that need company-wide protection can upgrade to connect AgentCloak Desktop with AgentCloak Server, an AI-native Rust container that deploys in any cloud, on-premises, or air-gapped environment. AgentCloak Server adds cloaking of uploaded files, richer digital twins that keep related values grouped across prompts, generalization of terms and numeric ranges, vertical data-protection packs for financial services, contracts, healthcare, and legal, and MCP data protection for Salesforce, HubSpot, and other apps.
With AgentCloak Server, IT departments can push a preconfigured AgentCloak Desktop with Intune and similar tools, collect audit logs, and use Microsoft Entra, Google Workspace, or Okta login. Cross-border data protection is also available with worldwide InCountry Sovereign Data proxies.
Availability
AgentCloak Desktop is available now at no charge. Install the browser extension or desktop app at agentcloak.ai/desktop.
Organizations interested in AgentCloak Server can contact sales@agentcloak.ai.
About AgentCloak InCountry
AgentCloak Sovereign AI and InCountry Sovereign Data protect customer data at every AI and human touchpoint and across borders. AgentCloak detects sensitive data using tuned AI models, substitutes values with synthetic and context preserving equivalents with a secure digital twin, and then restores the real values secured by identity context. AgentCloak is deeply integrated with leading enterprise products across the customer data lifecycle, including CDP, CIAM, CRM, customer service, and data lakes. AgentCloak InCountry is a CB100 2026 top private AI company, serves highly regulated Global 2000 companies, and has raised $50 million from leading investors.
View source version on businesswire.com:https://www.businesswire.com/news/home/20260918884045/en/
CONTACT: Media contact:
press@agentcloak.aihttps://agentcloak.ai
KEYWORD: UNITED STATES NORTH AMERICA CALIFORNIA
INDUSTRY KEYWORD: SOFTWARE TECHNOLOGY ARTIFICIAL INTELLIGENCE
SOURCE: AgentCloak InCountry
Copyright Business Wire 2026.
PUB: 09/18/2026 10:00 AM/DISC: 09/18/2026 10:00 AM
This text was published by aol.com and written by AOL. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us.
More in Enterprise & Industry
All →- Google adds unified sharing rules, AI Overviews, and Gemini integrations to Workspace · 1 src
- Huntress reports AI-driven rise in RMM abuse and cybercrime efficiency · 1 src
- AWS and ADP expand partnership to deliver AI-powered HR solutions · 1 src
- Spirit AI expects robot brains to hit ChatGPT-like breakthrough by next year · 2 src
- Microsoft patches CVSS 10.0 Azure AI Foundry flaw (CVE-2026-85889) · 1 src
Comments
via GitHub Discussions