Citizens trusted you with their data. AI tools are one paste away.
Across ministries, agencies, councils and public bodies, staff already use AI to draft letters, summarise case files and translate documents. The risk is what they paste into it: citizen records, ID numbers, case files, confidential documents. Tracehold checks what is sent to AI tools and, according to your policy, observes, warns, redacts or blocks. If nothing sensitive is found, the prompt goes through.
Tracehold catches it before it leaves the browser:
The extension checks the prompt and attached files before they are sent.
Event metadata (type of data, severity, destination, action) plus a masked sample of the value. Saving content is off by default.
Running in the browser extension, which recognises more than 300 AI-tool domains.
Built and hosted in the EU. The Security page explains what the console receives and where it runs.
Shadow AI doesn't respect departmental boundaries
Generative AI didn't arrive through a procurement process. It arrived through the browser, on every desk, in every agency, all at once. Helpful, fast, and one paste away from sending citizens' data to a service outside your controls. The intent is almost always innocent. The exposure isn't.
Citizen & case data
A caseworker pastes a full citizen record (name, national ID, address, benefits or health notes) into an AI tool to draft a reply. An official summarises an applicant's file. Each one is personal data leaving public control.
Confidential documents
Staff drop internal memos, draft legislation, procurement files and restricted reports into AI tools to "summarise this" or "tidy up the language." Sensitive policy and personal data go straight into a prompt.
Sprawl across agencies
Dozens of departments, agencies and local bodies adopt AI at their own pace with no central view, so nobody can answer the basic question: who is sending what to which AI tool?
Built to recognise the data a public body can't afford to lose
With more than 1,600 rules running in the browser extension, Tracehold recognises data that has a recognisable shape, such as identity numbers, IBANs, card numbers and credentials.
Citizen & personal data
National identity numbers from several countries (Ireland’s PPS, France’s NIR, the Dutch BSN, the Spanish DNI/NIE), IBANs and card numbers, validated with their check digit. Names on their own are not a finding, and health terms are flagged only when a personal identifier is close by. Free-form text needs the optional LLM classifier (Governance plan, beta).
Confidential documents
Documents carrying markings such as STRICTLY CONFIDENTIAL, TOP SECRET or DO NOT DISTRIBUTE, or your own keywords. A bare “CONFIDENTIAL” or “SECRET” is not a marking the engine recognises, and it does not read a memo to decide that it is confidential.
Credentials & keys
API keys, access tokens, passwords, private keys and connection strings hidden inside pasted code or config, caught before they reach an AI assistant and become someone else's.
Catches it at the source
Detection runs in the browser, before the prompt is sent to an AI tool: typed or pasted text, and files attached or uploaded, including text read from images by OCR in the browser. The console receives event metadata plus a masked sample of the value; saving content is off by default. The browser extension needs no proxy.
Warn, redact or block, based on your policy
Low-risk cases get a clear heads-up. The sensitive part can be stripped so the rest goes through. High-risk pastes are blocked outright, with a message that explains why, so staff learn, not just get stopped.
Covers the whole desk
Beyond the browser extension, the Windows agent covers USB drives, printing and screen captures of protected Office documents, and the Linux agent covers terminal AI tools such as Claude Code and Codex, so the protection follows the work, not just one tab.
Keywords you tell it to watch
Add your own keywords (agency and department names, confidential project codenames, named officials, key contractors) and Tracehold flags them when they appear in a prompt or attached file, on top of the built-in rules.
Built and hosted in the EU, with a clear account of what reaches us
For a public body, where the tool itself lives matters as much as what it does. Tracehold is built and hosted in the EU, and each organisation's data is isolated with Row-Level Security. In the browser extension, prompt text is not sent to the console by default.
Built and hosted in the EU
Tracehold is built and hosted in the EU. The Subprocessors page lists who processes data for us.
Optional features only
In the browser extension, prompt text is sent to a server only if your organisation enables the optional LLM classifier (Governance plan, beta; text in transit, not stored) or finding-content capture.
What the console receives
Event metadata (type of data, severity, destination, action) plus a masked sample of the value. Saving content is off by default.
Built and hosted in the EU
Each organisation's data is isolated with Row-Level Security. In a managed deployment the console receives event metadata plus a masked sample of the value, not the citizen record.
Uncontrolled AI use touches every framework a public body answers to
Public bodies sit under some of the strictest obligations in Europe. The same paste can trip more than one rulebook at once, and in Spain, the Esquema Nacional de Seguridad sets the bar on top.
GDPR
Pasting a citizen's personal data into a public AI tool can be an unlawful transfer to a third party. Checking it before the send helps keep that data under public control.
NIS2
Public administrations are in scope for stronger cyber-risk management and accountability. Demonstrable control over how data leaves to AI tools, backed by an audit trail, is part of showing you manage the risk.
ENS (Spain)
The Esquema Nacional de Seguridad is not one of the six frameworks the product maps, and it is not part of the compliance score. A basic legacy report is available only through the API and references some of its measures; it is not an assessment or a certification.
EU AI Act
As public services adopt AI, you need to know where and how it's used, and govern it. Governance maps findings to 8 controls, such as Art. 26 (AI systems in use by the deployer), Art. 12 (record keeping) and Art. 14 (human oversight).
Compliance mapping you can put in front of leadership and auditors
In the Governance plan, Tracehold maps findings to 39 controls across six frameworks (GDPR, the EU AI Act, DORA, NIS2, ISO 27001 and SOC 2) and exports a report for each, in PDF or CSV. It supports your audits; it is not legal advice or a certification.
Risk, not theatre
The mapping reflects what your people are actually doing with AI today: what's being caught, how often, how severe, not a point-in-time checklist that's stale by lunchtime.
Evidence on demand
Each finding is kept as a record of its type, severity, destination and action, with a masked sample of the value rather than the full text, so you can show an auditor how AI use is controlled without exposing the underlying data.
Support for your audits, not a certification
The mapping is a product judgement about which detections evidence which obligation. It supports your audits; it is not legal advice or a certification. Review it with your compliance team.
From "we think we're fine" to "here's the picture"
Most public bodies can't answer a simple question from leadership: how much citizen data is going to AI right now, across our agencies? Compliance mapping answers it and shows the trend as your controls take effect.
Reports for GDPR, the EU AI Act, DORA, NIS2, ISO 27001 and SOC 2 come with the Governance plan, in PDF or CSV. See compliance in depth →
No proxy, no network surgery.
A public body can't bolt a new inline proxy into its network on a whim. The browser extension needs no proxy and no network change, and you push it with the device management you already run.
Pushed via your MDM
The console generates the managed-policy files (.reg, .ps1, .json) and the MSI, and there is a guide for Microsoft Intune. Packages and policy come from the console; your MDM or GPO does the rollout.
Start observe-only
Watch and report without blocking anyone. You see what would have been caught and where your real exposure is across agencies before you turn on warnings, redaction or blocking.
Plugs into your stack
Sign-in with OIDC or LDAP/AD. Findings from the browser extension go to your SIEM over HTTPS (Splunk HEC or CEF), and alerts go to Slack, Microsoft Teams or your own webhook URL.
Start free, scale to governed
Begin with the free extension, add managed visibility and compliance reporting when you're ready.
Free
The free browser extension needs no account and is for one person. It has no console and no policies.
See what's includedInsight
Managed visibility into AI exposure across your agencies: the console, policies to block, warn or redact, and file scanning.
See InsightGovernance
Everything in Insight, plus compliance reports mapped to six frameworks, multi-tenant spaces and the LLM detection engine (beta).
Talk to our teamWhat public-sector buyers ask first
Does citizen data reach Tracehold?
Do we need a proxy or network changes?
Does Tracehold create a new sovereignty risk?
How does it help with GDPR, NIS2 and the EU AI Act?
Do you cover Spain’s Esquema Nacional de Seguridad?
Can it cover many agencies and departments at once?
Does Tracehold hold any certification?
Which plan do we need?
See what your organisation is already sending to AI
Start an observe-only pilot and get a clear, honest picture of where citizen data and confidential documents are leaking to AI tools across your agencies, with zero disruption to your people.