Leaders Insights
Leaders Insights

Stay at the top of your field, a little every day.

DomainsMarketingDataFinanceAI
ResourcesLearnTestToolsBlogGlossary
© 2026 Leaders Insights — All rights reserved.
Tracks/Data in the public sector/Governance, privacy and checks/Writing consent and data-sharing agreements that survive an audit
2/4+150 XP

Governance, privacy and checks

10The privacy laws that actually govern your data+15011Writing consent and data-sharing agreements that survive an audit+15012Running a privacy impact assessment before you launch+15013Preparing for a data audit without the scramble+150

Writing consent and data-sharing agreements that survive an audit

# Writing consent and data-sharing agreements that survive an audit

An auditor at a state workforce agency once flagged a single sentence in a data-sharing MOU: "data may be used to improve program outcomes." That phrase, vague and open-ended, triggered a finding that froze a $40 million SNAP-to-employment pilot for six months. The fix took one afternoon. The rewrite took a lawsuit-averse legal team two months to sign off. This lesson shows you how to write the sentence correctly the first time.

The scenario: your MOU

Imagine your state's Department of Health and Human Services (HHS agency, state-level) wants to share SNAP (Supplemental Nutrition Assistance Program) and TANF (Temporary Assistance for Needy Families) enrollment data with the state Department of Labor's workforce program. Goal: proactively enroll benefits recipients in job training.

You need a Memorandum of Understanding (MOU), the standard instrument for inter-agency data sharing, sometimes called a Data Sharing Agreement (DSA) or Interconnection Security Agreement (ISA) when systems, not just data, are linked.

This is the single riskiest document in a public data program, because it sits at the intersection of three legal regimes:

  • Federal: Privacy Act of 1974 (governs federal systems of records), and program-specific statutes like 7 CFR 272.1(c) for SNAP data confidentiality.
  • State: most states have their own public records and privacy statutes; some (like California's CCPA-adjacent public sector rules) impose extra layers.
  • Sector-specific: if health data is involved, HIPAA (Health Insurance Portability and Accountability Act) may apply even to a benefits agency if it touches Medicaid eligibility data.

The clause auditors flag #1: purpose limitation

Purpose limitation means data collected for one purpose (determining SNAP eligibility) cannot be reused for an unrelated purpose (workforce targeting) without explicit legal authority and consent language.

Auditors from the Office of Inspector General (OIG) or state comptroller offices specifically test this because it is the easiest violation to prove: they just read the MOU's "permitted uses" section and compare it to what the receiving agency actually did with the data.

Weak clause (fails audit):

> "Data shared under this agreement may be used by [Workforce Agency] to support program improvement and related initiatives."

Strong clause (survives audit):

> "Data shared under this agreement shall be used solely to: (a) identify SNAP recipients eligible for referral to the WIOA Title I Adult program under 20 CFR 680.110; (b) generate outreach lists limited to name, contact information, and eligibility category. Any use beyond (a) and (b) requires a written amendment executed by both parties' data governancedata governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.View full definition → officers."

The difference: named statute, named data fields, named use case, and an explicit amendment mechanism instead of an escape hatch.

The clause auditors flag #2: re-disclosure and downstream sharing

Re-disclosure is when the receiving agency shares the data with a third party (a contractor, a nonprofit partner, another agency). This is where most real-world breaches originate, not from hacking, but from a workforce agency handing a CSV file to a job-placement contractor without a matching agreement in place.

The Government Accountability Office (GAO) has repeatedly found this gap in audits of federal-state data sharing; see the GAO's reports on data sharing and privacy for real case examples.

Your MOU needs a flow-down clause: any obligation on the primary recipient extends automatically to subcontractors, with no re-disclosure allowed absent a signed sub-agreement mirroring the original MOU's terms.

Section 7 — Re-disclosure Prohibition
7.1 Recipient shall not disclose Shared Data to any third party,
    including contractors, without prior written consent of Discloser.
7.2 Any approved third party must execute a Data Use Agreement
    incorporating Sections 3 (Permitted Uses), 5 (Security Controls),
    and 9 (Breach Notification) of this MOU by reference.
7.3 Recipient remains liable for third-party non-compliance.

Consent: when do you actually need it?

Not every data share requires individual consent. Government-to-government sharing for a compatible statutory purpose often relies on a routine use exception (a Privacy Act term meaning the disclosure is compatible with the purpose for which the data was collected).

But if the workforce program plans to actively market services or share data with a private training vendor, many states require affirmative consent, often via a consent-to-release form signed at benefits intake.

Rule of thumb for your MOU drafting checklist:

1. Is there a specific statute authorizing this exact share? If yes, cite it, and routine use may suffice.

2. Does the share cross into a new program population or a private-sector partner? If yes, get affirmative consent.

3. Is any data sensitive category (health, immigration status, minors)? If yes, default to consent plus a heightened security review.

Security and audit-trail clauses

Auditors also test whether the MOU specifies technical safeguards, not just legal promises. Reference concrete standards:

  • NIST SP 800-53 (security controls catalog) for federal systems, free from NIST.
  • FedRAMP authorization if a cloud vendor is in the data path.
  • Encryption at rest and in transit (AES-256 is the common baseline as of 2026, stated here as an established industry standard, not a specific invented figure).

Your MOU should require:

  • A named data custodian on each side (a specific person, not a title alone) accountable for the data.
  • Logging of every query or extract against the shared dataset, retained for a minimum period matching your state's records retention schedule (commonly 3 to 7 years, varies by state).
  • A breach notification clause with a specific time window (72 hours is the GDPR standard in Europe; many US state agencies now mirror this even though the Privacy Act itself has no hard deadline).

Knowledge check

1. Why did the phrase 'data may be used to improve program outcomes' trigger an audit finding in the SNAP-to-employment MOU?

2. What is the core legal concern behind 'purpose limitation' in a data-sharing MOU?

3. Why does an MOU sharing SNAP enrollment data for workforce targeting sit at the intersection of multiple legal regimes rather than just one?

MULTIPLE CHOICE

4. Select ALL correct answers about why the workforce data-sharing scenario is legally complex.

Select all the correct answers.

MULTIPLE CHOICE

5. Select ALL correct answers about what an MOU/DSA needs to avoid an audit finding related to purpose limitation.

Select all the correct answers.

The retention and destruction clause

A shockingly common audit finding: the MOU never says when data must be deleted. Without an end date, "temporary" data shares become permanent shadow databases.

Write a sunset clause:

> "Shared Data shall be retained by Recipient no longer than 24 months from receipt, after which it shall be securely destroyed per NIST SP 800-88 media sanitization guidelines, with a certificate of destruction provided to Discloser within 30 days."

Pick a duration tied to the actual business need (a training program cohort cycle, an eligibility redetermination period), not an arbitrary round number.

A quick pre-signature audit checklist

Before either agency signs, run this five-point check, the same one most state auditors use:

1. Named statute authorizing the share (not "applicable law").

2. Enumerated data fields (not "relevant data").

3. Named permitted uses tied to a specific program citation.

4. Re-disclosure clause with flow-down obligations.

5. Retention and destruction with a specific timeline and method.

If any of these five is written in vague language, an auditor will flag it, and rightly so: vagueness is the technical mechanism by which scope creep happens.

🎬 [VIDEO: "Data Sharing Agreements Explained" - youtube.com - search for recent public-sector data governancedata governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.View full definition → webinars from NASCIO (National Association of State CIOs) covering MOU best practices]

Key Takeaways

  • Purpose limitation clauses must name the statute, the data fields, and the exact use case. Vague phrases like "program improvement" are the single most common audit finding.
  • Re-disclosure requires a flow-down clause. Any contractor or third party touching shared data needs a mirrored sub-agreement; the primary recipient stays liable.
  • Consent is not always required, but you must justify why not. Cite the specific "routine use" or statutory exception if you're skipping individual consent.
  • Security and retention clauses need concrete numbers: encryption standards, breach notification windows (72 hours is a common benchmark), and a hard data destruction date, not an open-ended "as needed."
  • Run the five-point pre-signature checklist (statute, fields, uses, re-disclosure, retention) before any MOU goes to signature; it mirrors what GAO and state auditors actually test.

Previous

The privacy laws that actually govern your data

Next

Running a privacy impact assessment before you launch