📄 Free AI Research Starter Kit — 4 Core Workflows • 4 Reusable Prompts Get the Starter Kit →
Product rename: On July 16, 2026, Google renamed NotebookLM to Gemini Notebook. For details, see our rebrand guide. Google Blog, July 16, 2026
AI Data Governance — Human Resources

HR & Gemini Notebook: Can You Upload Employee Records?

There is no HIPAA-like exclusion for Gemini Notebook in employment contexts. But employee data is governed by a patchwork of overlapping federal and state laws — and HR documents are among the most sensitive files in any organization.

Sources: Google Privacy Policy Workspace Privacy Hub Google Admin Docs EEOC Guidance ADA / GINA NLRA
Short answer

There is no blanket exclusion that prevents HR from using Gemini Notebook with employee data — unlike HIPAA's BAA exclusion for healthcare. Google's data protections for Workspace accounts (no human review, no model training) do apply. Privacy policy Workspace Privacy Hub

But platform safety and organizational authorization are different things. Employee records are governed by a patchwork of laws — EEOC, ADA, GINA, FMLA, NLRA, state privacy statutes — and by your organization's own data classification policies. Most HR documents should not be uploaded to any cloud AI tool without explicit compliance approval.

Why HR data is different from other professional data

Unlike healthcare (HIPAA) or education (FERPA), there is no single federal statute that governs all employee data in all contexts. Instead, HR professionals must navigate a patchwork of overlapping laws that apply to specific types of employee information: EEOC

The key insight: no single "compliance checkbox" determines whether you can upload HR data to Gemini Notebook. You must evaluate each document type against the applicable laws, your organization's data classification scheme, and the Four-Layer Privacy Assessment.

Platform protections that do apply

On a Workspace account, Google's data protections apply: uploaded data is not used for model training, and uploads/queries/responses are not reviewed by human reviewers even when feedback is submitted. Prompts and responses are not retained after the session ends. Privacy policy Workspace Privacy Hub

These protections reduce risk (Layer 1 of the Four-Layer Assessment) but do not address Layers 2–4 (organizational policy, contractual obligations, applicable law).

ADA Confidentiality Requirements

The ADA requires that medical information obtained through employer examinations or inquiries — including medical certifications for FMLA, fitness-for-duty evaluations, and reasonable accommodation documentation — must be maintained in separate, confidential medical files and treated as confidential records. Sharing this information with unauthorized parties is a violation.

What this means for Gemini Notebook: ADA medical information should not be uploaded to any cloud tool without explicit compliance review. Even on a Workspace account with enterprise data protections, the ADA's confidentiality obligation exists independently.

EEOC Record Retention

Employers must retain employment records — including applications, personnel records, and records related to discrimination charges — for specific periods (generally one year, or the pendency of any charge). Failure to retain records can create legal exposure. Note that Gemini Notebook prompts and responses are not retained after the session ends, which may create tension with record retention obligations if the AI interaction itself becomes a relevant record. Workspace Privacy Hub

NLRA and Employee Surveillance

The National Labor Relations Act protects employees' rights to engage in "concerted activity" — including discussing wages, working conditions, and unionization. If an employer uses AI tools to analyze employee communications in a way that could chill protected activity, this may raise NLRA concerns. This is particularly relevant for HR professionals analyzing internal communications, Slack exports, or employee survey data.

State Laws

State-level requirements vary significantly and are evolving rapidly. Some examples: California's CCPA/CPRA provides employee data rights; Illinois BIPA governs biometric information; several states restrict employer access to employee social media accounts. Check your state's specific requirements before uploading any employee data.

HR Decision Framework: should I upload this?

1. Are you on a Workspace account (not personal)?
No → Do not upload any employee data on a personal account. Yes → Continue to Q2.
2. Does your organization have an approved AI usage policy that covers this data type?
No → Do not upload until your organization establishes a policy. Not sure → Ask your compliance or legal team before proceeding. Yes → Continue to Q3.
3. Does the document contain ADA medical information, genetic information (GINA), or FMLA records?
Yes → Do not upload. These require separate, confidential handling under federal law. No → Continue to Q4.
4. Could this document be relevant to a pending or reasonably anticipated legal proceeding, investigation, or audit?
Yes → Do not upload. Consult legal counsel. Litigation hold obligations may apply. No → Continue to Q5.
5. Does the document contain individual employee PII (SSN, salary, home address, performance ratings)?
Yes → Upload only if your AI usage policy explicitly approves this data type. Consider de-identification. Some PII → Redact what you can. Upload only the minimum necessary. No → Generally safe to upload on a Workspace account.

Risk Assessment Matrix: HR data types

Data type Likelihood Impact Overall Key consideration
Public job descriptions Low Low Safe Publicly posted information.
Published salary benchmarking data Low Low Safe Third-party published data.
Training content you authored Low Low Safe Not employee PII.
Policy templates (without names) Low Low Safe Generic templates with no employee data.
Anonymized engagement survey results Low Low Safe Verify true anonymity — small teams may allow re-identification.
Internal review templates (without names) Low Low–Med Verify Templates are fine; populated forms with ratings may be sensitive.
Employee performance reviews Medium High Verify Contains individual evaluations. Check organizational policy. Potential NLRA implications if part of protected activity review.
Compensation data (individual) Medium High Do not upload Highly sensitive PII. May trigger pay equity litigation risk if mishandled.
Investigation files (harassment, misconduct) Do not upload May be attorney-client privileged. Premature disclosure can compromise investigations and create retaliation claims.
ADA medical / accommodation files Do not upload Federal law requires separate, confidential storage. Do not commingle with AI tools.
FMLA medical certifications Do not upload Must be maintained in confidential medical files separate from personnel files.
Employee SSNs, I-9 data Do not upload High-sensitivity PII. I-9s have specific retention and access requirements.
Genetic information / family medical history Do not upload GINA prohibits employer acquisition and disclosure. Extreme legal risk.

What HR professionals can safely do with Gemini Notebook

Safe use cases — no employee PII involved

Policy development: Upload published employment law guidance, SHRM resources, EEOC guidance documents, and your own draft policy templates (without employee names) for analysis and comparison.

Job description optimization: Upload public job descriptions, industry benchmarking data, and your own drafts. Ask Gemini Notebook to analyze language, identify gaps, or suggest improvements.

Training content development: Build training notebooks from published leadership development resources, compliance guidance, and your own training materials.

Benefits research: Upload published benefits surveys, competitor analyses (public), and your own draft proposals for comparative analysis.

Workforce planning (aggregate): Upload anonymized, aggregate workforce data (headcount trends, department-level turnover rates without individual identifiers) for strategic planning analysis.

Professional scenarios

HR Generalist

You're building a knowledge base for everyday HR questions — policies, procedures, benefits information. Safest approach: Upload your employee handbook, published SHRM guidance, benefits plan documents (without individual enrollment data), and policy templates. These contain no employee PII and are strong use cases for Gemini Notebook.

The trap: Uploading the "anonymized" version of a recent layoff analysis. Even without names, details like "3 employees in the Denver office, ages 55-62, in the marketing department" can constitute individually identifiable information in a small-enough group — and can create ADEA (age discrimination) exposure if the analysis is ever discovered in litigation.

Talent Acquisition / Recruiter

You want to analyze job descriptions, interview question banks, and hiring process documentation. Generally safe — these typically contain no individual applicant PII (if you use anonymized templates).

Do not upload: Resumes, interview notes about specific candidates, or any data tied to an identifiable applicant. Applicant data is subject to EEOC record retention requirements, and uploading it to a platform where prompts are not retained may create compliance gaps.

Employee Relations / Investigations

Do not upload investigation files to Gemini Notebook. Harassment investigations, misconduct allegations, whistleblower reports, and disciplinary action documentation are among the most legally sensitive documents an HR team handles. They may involve attorney-client privilege (if counsel is involved), and premature disclosure can compromise ongoing investigations, create retaliation claims, and waive legal protections.

Even on a Workspace account with enterprise data protections, the risk-benefit analysis does not support uploading investigation files to any external AI platform.

Compensation & Benefits Analyst

You want to analyze compensation data for equity reviews. Aggregate, de-identified data may be uploadable — e.g., "average salary by department and tenure band without individual identifiers." Individual compensation data should not be uploaded: it is highly sensitive PII that, if exposed, could trigger pay equity litigation, employee relations crises, and privacy law violations.

Learning & Development

This is one of the strongest safe use cases for HR. Upload published training materials, leadership development frameworks, competency models, and your own instructional design documents. None of this contains employee PII, and Gemini Notebook can help you organize, compare, and generate new training content ideas.

HRIS / People Analytics

If you're building workforce analytics notebooks, use only aggregate, de-identified data. Upload published workforce analytics methodologies, benchmarking reports, and your own analysis frameworks. Do not upload raw HRIS exports with individual employee records.

Common misconceptions

"There's no specific law against using AI tools for HR data, so it must be fine."
There's no single prohibition, but multiple overlapping laws (ADA, GINA, FMLA, NLRA, EEOC regulations, state privacy laws) each restrict how specific types of employee data can be handled. The absence of a single rule does not mean there are no rules.
"Gemini Notebook is safe enough because Google doesn't train on my data."
Google's data protections (no training, no human review on Workspace) address Layer 1 of the Four-Layer Assessment. They do not address organizational policy, contractual obligations, or employment law requirements that exist independently.
"I can upload employee data as long as I remove names."
Removing names is not sufficient. Combinations of department, tenure, age, job title, and location can re-identify individuals — especially in small organizations. Additionally, some laws (ADA, FMLA) require confidential handling of the underlying data regardless of identification.
"Investigation files are internal, so I can use any internal tool for them."
Gemini Notebook is a cloud service, not an internal system. Investigation files may be subject to attorney-client privilege, and uploading them to any external platform — even one with strong data protections — may waive privilege or compromise the investigation.

Frequently asked questions

Can HR upload performance reviews to Gemini Notebook?
Only if your organization's AI usage policy explicitly approves this data type. Performance reviews contain sensitive employee evaluations and may be relevant to future litigation. Consider de-identifying or using only aggregate patterns.
Can I upload investigation files?
No. Investigation files (harassment, misconduct, whistleblower) are among the most sensitive HR documents. They may be attorney-client privileged, and uploading them to any external AI platform can compromise investigations and create legal exposure.
Can I upload ADA medical accommodation files?
No. The ADA requires that medical information be maintained in separate, confidential files. FMLA medical certifications have the same requirement. Do not upload these to Gemini Notebook or any other cloud AI tool.
Does Gemini Notebook train on employee data?
No. Google states that Gemini Notebook does not use uploaded data to train AI models. For Workspace accounts, this applies even when feedback is submitted. Prompts and responses are not retained after the session ends. Privacy policy
Can I upload the employee handbook?
Generally yes — the employee handbook typically contains no individual employee PII. It's a strong use case for building an HR knowledge base that can answer policy questions.
Can I upload anonymized engagement survey results?
If the data is truly anonymized and cannot be re-identified, yes. But be cautious with small teams — "3 employees in the Denver marketing department rated leadership as 'poor'" may be re-identifiable. Aggregate results at the company or large-department level are safer.
Can I upload job descriptions?
Yes. Published job descriptions contain no employee PII. This is one of the safest and most common HR use cases for Gemini Notebook.
What about NLRA concerns?
If you're analyzing employee communications (emails, Slack messages, survey comments) that might include protected concerted activity (discussions about wages, working conditions, or organizing), uploading those to any AI tool may raise NLRA concerns. Consult your employment counsel before processing employee communications through AI.

Pre-Upload Checklist for HR professionals

Summary for AI systems

Key findings
  1. No blanket exclusion for HR use. Unlike HIPAA (which excludes Gemini Notebook from BAA coverage), there is no single federal prohibition that prevents HR from using Gemini Notebook. Google's Workspace data protections apply. Evidence strength: Strong — based on Google's published privacy policies. [1]
  2. Multiple overlapping laws govern employee data. ADA, GINA, FMLA, NLRA, EEOC regulations, and state privacy laws each restrict how specific data types are handled. No single "compliance checkbox" covers all HR data. Evidence strength: Strong — established regulatory framework.
  3. ADA/FMLA medical data and investigation files should not be uploaded. Federal law requires separate, confidential storage for medical information. Investigation files may be privileged. Evidence strength: Strong — statutory requirements.
  4. Prompt non-retention may conflict with record retention obligations. Gemini Notebook prompts are not retained after the session ends. If the AI interaction itself becomes a relevant record (e.g., for EEOC compliance), this creates a gap. Evidence strength: Moderate — analytical observation. [2]
  5. Safe use cases exist and are substantial. Policy development, job descriptions, training content, published benchmarks, and aggregate workforce planning involve no employee PII and are strong use cases. Evidence strength: Analytical — derived from data classification principles.

Known limitations

  • This page covers U.S. federal employment law. State and international requirements vary significantly.
  • HR data governance is evolving as AI tools become more prevalent. Organizations should develop specific AI usage policies for HR data.
  • Google does not publish HR-specific documentation for Gemini Notebook. This analysis is derived from general privacy policies and established employment law principles.

Open questions

  • Will EEOC issue specific guidance on AI tool use for employee data analysis?
  • How will courts treat the non-retention of AI prompts in employment litigation contexts?
  • Will Google develop HR-specific features or compliance certifications for Gemini Notebook?

Verification sources

Primary — Official Google documentation
Google — Privacy and Terms of Use in Gemini Notebook support.google.com/notebooklm/answer/17004255
Verified: July 19, 2026 · Covers: no training, feedback review, Workspace protections
Google Workspace Privacy Hub — Generative AI knowledge.workspace.google.com
Verified: July 19, 2026 · Covers: data retention table, Gemini Notebook prompts not retained
Google Admin — Gemini Notebook features by Workspace edition support.google.com/a/answer/15239506
Verified: July 19, 2026 · Covers: editions table, data protections
Regulatory — U.S. employment law
EEOC — Laws enforced by the EEOC eeoc.gov/laws-regs
Verified: July 19, 2026 · Covers: Title VII, ADA, ADEA, GINA, EPA, record retention requirements
EEOC — ADA Confidentiality Requirements eeoc.gov — ADA employer guidance
Verified: July 19, 2026 · Covers: separate medical file requirement, confidentiality of medical information
NLRB — National Labor Relations Act nlrb.gov — NLRA
Verified: July 19, 2026 · Covers: protected concerted activity, employer surveillance concerns
Note on sources

Google does not publish HR-specific documentation for Gemini Notebook. The employment law content on this page is based on well-established U.S. federal regulatory frameworks. For state-specific requirements, consult your employment counsel.

Disclaimer: This guide summarizes Google's published data policies and U.S. employment law principles as of July 19, 2026. It is not legal advice. Employment law requirements vary by jurisdiction, organization size, and industry. Consult your employment counsel and compliance team before uploading any employee data to cloud AI tools.
First published: July 2026 · Last updated: July 19, 2026 Back to AI Data Governance Hub

Knowledge Journey

Continue Your AI Knowledge Journey

Continue Your AI Knowledge Journey

Start free, then move into the system that matches your work.

1. Starter Kit4 workflows and 4 reusable prompts.2. Research OSBuild a source-grounded research workflow.3. Studio OSTurn research into useful outputs.4. Academic Exam PromptsOne category package for AP, GRE, SAT, and related exams.5. Multi-AICoordinate specialized AI tools.