HR & Recruiting
Use AI in HR and Recruiting Without Exposing Candidate and Employee Data
What the employee types
Draft a reply to Ada Lovelace about her
refund of CHF 2,400 to card
4532 7781 0021 9047
IBAN CH93 0076 2011 6238 5295 7
What the AI provider receives
Draft a reply to mf_mask_170b about her
refund of mf_mask_9c2e to card
mf_mask_41af
IBAN mf_mask_d7e0
574
Built-in detectors for personal data and secrets
47
Country packs, checksum-validated where the format has one
9
AI providers behind one gateway, plus the desktop agent and proxy
Where people data meets AI in HR
A recruiter pastes three CVs into a chatbot and asks for interview questions. Each has a name, a home address, a phone number, a date of birth in some countries, and sometimes a photo. An HR business partner drafts a letter about a long sickness absence, and pastes the case notes to get the tone right. A manager rewrites a performance review.
HR data is some of the most sensitive in any company. Case notes can reveal health, union membership, religion or family circumstances, which are special-category data under GDPR Article 9. In many countries, employee data is also subject to employment law and works council rights. Staff often use personal AI accounts for this work simply because nobody has given them an approved route.
Maskflare gives them that route: the desktop agent and browser protection for HR staff, the gateway for any assistant you build, and warn rules where a person should stop and think before sending. Decisions about hiring and people stay with people, and the AI Act's rules on those systems are a separate assessment.
Where hr & recruiting is actually exposed
5 failure modes specific to this environment, not a restatement of general security advice.
01
CVs pasted or uploaded for screening help
A CV carries contact details and a career history. The desktop agent masks identifiers before the prompt reaches ChatGPT, Claude, Gemini, DeepSeek, Perplexity, Mistral Le Chat and Microsoft 365 Copilot. PDF and Word uploads are scanned under your document policy and recorded or refused, not rewritten, and uploads to a separate storage host are refused. Plain-text files are masked in place where they're readable in the request.
02
HR case notes with sensitive details
Absence, grievance and disciplinary notes can reveal health, beliefs or union membership. The special categories pack can flag this free text, and a warn rule asks the person to confirm or give a reason before it leaves.
03
Performance reviews and salary discussions
Reviews are tied to a named person, and pay data appears next to it. Names and IDs are masked, while a salary figure isn't an identifier, so sensitive pay work needs a custom rule or a policy on approved tools.
04
Personal AI accounts
HR staff may turn to a personal account when no approved route exists. AI app discovery shows which apps are used on managed devices, and policy can limit personal ChatGPT and Claude accounts.
05
Recruiting tools built on third-party models
If your team builds a sourcing or drafting assistant, candidate data flows to the model by default. The gateway applies the same rules through a base-URL change, and embeddings use deterministic tokens.
Who this page is for
- Head of People
- Wants the team to save time with AI without breaking employees' trust.
- Talent acquisition lead
- Needs recruiters to work faster on candidate material without sending CVs to a provider in clear text.
- Data protection officer
- Has to document transfers of employee data, and special-category data in particular.
- Works council or employee representatives
- May have rights over monitoring tools, and will ask what's recorded about individuals.
- Legal and compliance
- Needs AI that influences hiring or employment assessed separately under the EU AI Act.
How a team would actually run this
Giving HR an approved way to use AI
- 01 Bring in employee representatives early where they have rights over monitoring tools, and explain that Maskflare records metadata and never prompts.
- 02 Deploy the desktop agent to HR and recruiting with Intune, Jamf or Group Policy. Only AI hosts are routed through it.
- 03 Apply the GDPR profile with your country packs. The default starter masks the special categories pack; set it to alert first if you want to see what turns up.
- 04 Set warn rules for special categories, so people give a reason or choose to mask before sending, and mask everyday identifiers by default.
Summarizing a case without exposing the person
- 01 The HR partner pastes case notes and asks for a neutral summary.
- 02 If a warn rule matches, the desktop app holds the message, names the kinds of data found, and offers mask and send, send with a reason, or cancel.
- 03 The provider receives tokens instead of identifiers. The summary may show tokens where it refers to people, and the partner fills them in or keeps them.
- 04 The console records the category, the app and the choice, never the text, so a reviewer can see patterns without reading the case.
Which modules apply, and why
Each links to the module page, where the boundaries and development status are set out in full. Or start at the Maskflare platform overview.
Detection engine
Maskflare Detection Engine
Detects and masks names, contact details, national IDs and bank details before a prompt reaches an AI provider, and on API traffic restores them in the answer.
For your people
Maskflare User Coaching
A warn rule asks HR staff to confirm or justify before sensitive categories leave, on managed devices.
For your people
Maskflare Workforce
The desktop agent covers HR and recruiting teams on AI chat sites, deployed through device management.
For your people
Maskflare AI App Discovery & Control
Shows which AI apps are used on managed devices, as metadata only, including ones no one approved.
Detection engine
Maskflare File Security
Scans CVs and documents under your policy, and masks plain-text files in place.
For your apps
Maskflare AI Gateway
Puts recruiting assistants you build behind the same rules with a base-URL change.
The data, and how it's found
Typical data in this sector, the detectors that cover it, and where you'll need your own rules.
| Data | How Maskflare finds it | Profile or rule |
|---|---|---|
| Names and contact details | Person names pack, local named-entity layer, email and phone detectors | GDPR, UK GDPR, US PII |
| Dates of birth | Personal date detectors | GDPR, UK GDPR, US PII |
| National and social security IDs | 47 country packs with checksum and format validation | GDPR, US PII |
| Bank details for payroll | IBAN validation by checksum | GDPR, UK GDPR |
| Health, beliefs and union membership in free text | Special categories pack: masked on the default starter, alert on Monitor first | Your policy decision |
| Salaries and internal employee IDs | No built-in detector: custom rules or Exact Data Match | Your own rules |
| Photos on CVs | Image policy: allow, block or scan text; faces aren't detected | Your policy decision |
Regulatory context
Readiness and relevance, not certification. Nothing here is a claim to hold an audit or authorization we do not have.
- GDPR Arts. 9 and 88
- Article 9 covers special-category data, and Article 88 lets member states set more specific rules for employment. Reversible tokens remain pseudonymized personal data for you, so your legal basis still applies.
- EU AI Act, Annex III
- AI used for recruitment and worker management is listed as high-risk. Regulation (EU) 2026/1744 moved those obligations to 2 December 2027. Maskflare doesn't assess or classify AI systems, and masking doesn't take a hiring tool out of scope.
- EU AI Act, Art. 5
- Bans emotion recognition in the workplace, with narrow exceptions for medical or safety reasons. That's about what an AI system does, and Maskflare has no bearing on it.
- Works council rights
- In countries such as Germany, tools that can monitor employees may need works council involvement. Maskflare records metadata such as category, app and user, so check this before rollout.
- Swiss FADP and Code of Obligations Art. 328b
- Limit employers' processing of employee data to what relates to the job. Masking supports data minimization, and the legal assessment stays with you.
What Maskflare does not do
- It doesn't screen, rank or score anyone, and masking a CV doesn't make an AI hiring process fair or lawful.
- Removing a name doesn't remove identity. A job title at a small company can still point to one person, and masking doesn't remove bias signals such as career gaps.
- Faces in CV photos aren't detected. Use your image policy if photos are a concern.
- Coaching dialogs need the desktop agent on managed devices, so personal phones are out of reach.
- Documents, including scanned pages read with OCR, are scanned under your document policy and recorded or refused, not rewritten. In browser apps, picture uploads pass unread or are refused, and uploads that go to a separate storage host are refused.
- SOC 2 Type II and ISO 27001 audits are in progress.
5 questions to ask any vendor here
Including us. If our answer is worse than someone else's, you should know that before you buy, not after.
- 01Does the provider ever receive a candidate's name or an employee's ID, or only tokens?
- 02What exactly is recorded about each employee using the tool, and who can see it?
- 03Can people give a reason, or choose to mask, instead of being blocked?
- 04How does it handle CVs as PDF or Word files, and photos on them?
- 05Does it make or influence any decision about a person?
Workflow guides for this industry
How a team applies each of these, including the contrast with the tool it replaces.
Definitions worth agreeing on first
HR & Recruiting questions we get asked
Can HR teams paste CVs into ChatGPT?
With the desktop agent installed, names, contact details and IDs are masked before the text reaches ChatGPT. The answer may show tokens where it refers to them, and the person fills those in or keeps them. Whether you may do this at all depends on your legal basis and candidate notices, so check with your DPO. This is not legal advice.
Is a recruiting chatbot high-risk under the EU AI Act?
AI used to recruit or select people, or to manage workers, is listed in Annex III as high-risk, and those obligations now apply from 2 December 2027. Whether your use falls within it needs a legal assessment. Maskflare doesn't make that assessment.
Does Maskflare monitor what employees type?
It never stores prompts or responses. It records metadata such as the app, the user, the time and the categories found. That can still count as a monitoring tool in some countries, so involve employee representatives where they have rights.
Can it detect health or union details in case notes?
The special categories pack finds these in free text. The default Protect now starter masks them, and the Monitor-first starter only counts them. It works from word lists and code lists, so it won't catch every indirect reference, and people still need to use judgment.
Does masking remove bias from AI screening?
No. Masking hides identifiers from a provider, and it isn't a fairness control. Career gaps, schools and wording can still act as proxies, and decisions need human review.
Where is Maskflare hosted?
In the EU by default, with on-premises deployment available for enterprise. Prompts and responses are never stored; masked values are kept encrypted for 30 days so answers can be restored.
See it on your own data
Start a free trial or talk to us with the identifiers and AI workflows your hr & recruiting team actually handles. Bring the case that makes it hard, because that is the part worth testing.
See it on your own data
Try Maskflare free.
Bring your hardest prompts.
28-day free trial or 10,000 requests, whichever comes first. Point one app at the gateway or enroll a few laptops,
and watch what gets masked. Or talk to us first.
- SOC 2 Type II: In progress
- ISO 27001: In progress
- EU hosting by default · on-premises available
- Prompts and files never stored
- Never used to train AI models
- GDPR · nDSG · DPA available
- Trust center →