Compliance & Ethics

What Is a Clawback Agreement? FRE 502(d) Explained

Most clawback clauses in federal protective orders won’t survive a contested motion. The difference is two paragraphs of drafting at the Rule 26(f) conference.

Alexander Cohan, Ph.D.

Alexander Cohan, Ph.D.

Legal technology researcher and data scientist specializing in AI governance for litigation teams. Expertise in NLP and AI-assisted document review.

Clawback agreement under FRE 502(d): privileged document being recalled before opposing counsel review
"Setting the Scene"

What a Clawback Agreement Actually Does

What is a clawback agreement? It’s the paper that says if a privileged document slips into a production by mistake, the producing party can demand it back and the receiving party has to return, sequester, or destroy it. The mechanics are straightforward. The legal effect is anything but. Whether that paper actually preserves the attorney-client privilege depends on which subsection of Federal Rule of Evidence 502 it rides on.

There are three tracks. Rule 502(b) is the default rescue valve. If the production was inadvertent, if the holder took reasonable steps, and if the holder promptly moved to rectify, no waiver. Rule 502(d) is the court order, and it’s the strongest of the three because the federal court can declare that the disclosure isn’t a waiver, full stop, including in any other federal or state proceeding. Rule 502(e) is the bare party agreement. It binds only the parties who signed it, which matters when third parties come knocking.

A small-firm litigator with a million-document custodial set knows the math. No privilege review catches everything. That’s why the right clawback structure isn’t optional. It’s the difference between an inconvenient phone call to opposing counsel and a privilege waiver you’ll have to defend in front of a judge asking whether your screening process counts as reasonable steps. Modern e-discovery practice assumes you’ll have something in place; federal magistrates assume you’ve read FRE 502 before showing up to the Rule 26(f) conference. There’s also a malpractice framing here that we’ll get to in Section IV, courtesy of former Magistrate Judge Andrew Peck. For now, hold this thought: a private agreement isn’t the same thing as a court order, and the difference is structural. The cost of getting it wrong runs through fee disputes, privilege exposure, and, in a bad scenario, a bar complaint.

"The Three Tracks"

FRE 502(b), (d), and (e): What’s the Difference?

The three subsections share a goal, cutting the cost and risk of privilege review in large productions, but operate on different machinery. Read the text closely. The words do a lot of work.

Rule 502(b) is the default safety net. Disclosure “does not operate as a waiver in a federal or state proceeding if” the disclosure is inadvertent, the holder took reasonable steps to prevent it, and the holder promptly took reasonable steps to rectify the error, including, if applicable, following Federal Rule of Civil Procedure 26(b)(5)(B). Three prongs, all of which the producing party has to satisfy. If your screening process can’t survive scrutiny, you don’t get the shield. The Advisory Committee called this the middle ground between strict-accountability and intent-only approaches.

Rule 502(d) is the court order. A federal court may order that the privilege or protection is not waived by disclosure connected with the litigation pending before the court, and that order also bars waiver in any other federal or state proceeding. Two clauses, both load-bearing. The court can enter the order by stipulation or on its own motion. And the protection travels: a 502(d) order entered in your federal case is enforceable in the state-court matter your adversary files three years later. The Advisory Committee Note frames the order as the mechanism by which courts authorize return of documents “irrespective of the care taken by the disclosing party.” That phrase will reappear in Section V.

Rule 502(e) is the bare party agreement. An agreement on the effect of disclosure in a federal proceeding is binding only on the parties to the agreement, unless it is incorporated into a court order. That last clause is the trap. A 502(e) agreement is contractual. It binds A and B. It doesn’t bind C, the regulator, the future plaintiff, or the state court. The Advisory Committee was explicit: if parties want protection against non-parties from a finding of waiver by disclosure, the agreement must be made part of a court order. Rule 26(b)(5)(B) walks through clawback mechanics (notify, sequester, return), but procedure can’t help when a spoliation fight in a parallel proceeding puts the same documents on the table.

"Reasonable Steps"

The 502(b) Test: How Courts Decide When Inadvertent Means Waived

Without a 502(d) order, you’re stuck arguing 502(b). And 502(b) is a hard test to pass when the production is large and the screening was sloppy. Three prongs, every one of them litigated.

Prong one: was the disclosure inadvertent? Courts read this narrowly. Coburn Group, LLC v. Whitecap Advisors LLC, 640 F. Supp. 2d 1032, 1038 (N.D. Ill. 2009), treats inadvertence as “much simpler” than reasonable steps, asking whether the party intended a privileged or work-product protected document to be produced or whether the production was a mistake. That part isn’t usually where producing parties lose.

Prong two is. “Reasonable steps to prevent disclosure” gets unpacked through a five-factor analysis that the Advisory Committee Note traces to Lois Sportswear, U.S.A., Inc. v. Levi Strauss & Co., 104 F.R.D. 103 (S.D.N.Y. 1985), and Hartford Fire Insurance Co. v. Garvey, 109 F.R.D. 323 (N.D. Cal. 1985): the reasonableness of precautions, the time taken to rectify, the scope of discovery, the extent of disclosure, and the overriding issue of fairness. Mt. Hawley Insurance Co. v. Felman Production, Inc. applied that framework and held the producing party hadn’t taken reasonable precautions. Mt. Hawley, 271 F.R.D. 125, 135 (S.D. W. Va. 2010). The lesson generalized fast. Rhoads Industries applied the analysis to an inadvertent production of over 800 privileged documents and found waiver as to documents not logged by the privilege-log deadline, preserving privilege only for the timely-logged subset. Rhoads Indus., 254 F.R.D. 216 (E.D. Pa. 2008). In Inhalation Plastics, Inc. v. Medex Cardio-Pulmonary, Inc., No. 2:07-CV-116, 2012 WL 3731483, at *6 (S.D. Ohio Aug. 28, 2012), the court found waiver as to 347 in-house counsel emails after concluding that the producing party had not taken reasonable precautions, the number of documents disclosed was significant, no privilege log accompanied the production, and the mitigation effort was thin.

Prong three is the rectification clock. The producing party has to move quickly under Rule 26(b)(5)(B), and courts look at how prompt is prompt. Two days is fine. Two months is not.

What actually counts as “reasonable steps”? The Advisory Committee Note, anticipating technology-assisted review by a decade: a party that uses advanced analytical software applications and linguistic tools in screening for privilege and work product may be found to have taken reasonable steps, depending on the circumstances. Analytical tools can satisfy reasonable steps. They don’t guarantee it. The record you build is what carries you through.

"Get Out of Jail Free"

Why 502(d) Orders Beat Bare Agreements

The reason federal practitioners routinely move for a 502(d) order at the Rule 26(f) conference, and why The Sedona Conference recommends one as a matter of course, isn’t theoretical. The 502(d) order can kill the 502(b) reasonableness inquiry when drafted properly, and it extends the protection to non-parties and future proceedings. The bare 502(e) agreement does neither.

Former Magistrate Judge Andrew Peck (S.D.N.Y., ret.) has been the most consistent voice on this point, and his formulation has been picked up by The Sedona Conference and reproduced widely in practitioner commentary.

That’s a strong opinion from one prominent bench voice. It connects directly to the malpractice exposure that follows lawyers using AI tools: when the cost of getting a 502(d) order is two paragraphs at the scheduling conference, failing to ask for one is hard to defend after a privilege waiver becomes public. The framing scales with volume. For a 500-document production, the calculus is different. For modern federal litigation, it isn’t.

The case law backs the framing. In Brookfield Asset Management, Magistrate Judge Maas held that even if AIG had “dropped the ball” on its privilege review, the parties’ 502(d) stipulation entitled AIG to claw back the disputed minutes no matter what the circumstances giving rise to their production were. Brookfield, No. 09 Civ. 8285, 2013 WL 142503, at *2 (S.D.N.Y. Jan. 7, 2013). In Rajala, Magistrate Judge Waxse enforced an earlier 502(d) order over the defendant’s complaint about a “document dump,” holding that the terms of the protective order, not the default provisions of FRE 502, governed inadvertent productions in the case. Rajala v. McGuire Woods LLP, No. 08-2638-CM-DJW, 2013 WL 50200, at *5 (D. Kan. Jan. 3, 2013). The earlier Rajala opinion explained why the court would enter the order over party objection: clawback provisions permit expeditious discovery without time-consuming and costly pre-production privilege reviews.

That cost reduction is the proportionality story. A 502(d) order lets you produce faster and cheaper because you don’t have to run privilege review to the standard required to survive a 502(b) reasonableness fight if something slips through. The order replaces that fight with a contractual return, and it does so in a way that respects the proportionality concerns Rule 26 was designed to address. The court-order shield also reaches third parties. Once the federal court enters a 502(d) order, the disclosure is also not a waiver in any other federal or state proceeding. That portability is the structural advantage 502(e) can’t replicate.

"Drafting Traps"

“Reasonable Steps” vs. “Irrespective of Care”

Here’s where careful drafting matters. A bare 502(d) order that just says “any inadvertent production is non-waiver” might not actually displace the 502(b) reasonableness inquiry. The cases are sharper on this than most practitioners realize.

The leading decision is Irth Solutions, LLC v. Windstream Communications LLC. The court surveyed three approaches to clawback agreements: a no-fault approach (return privileged documents on demand, full stop), a recklessness approach (clawback fails only if the production was completely reckless), and a default approach where Rule 502(b)’s reasonableness analysis applies unless the agreement displaces it. Irth Sols., No. 2:16-CV-219, 2017 WL 3276021, at *5 (S.D. Ohio Aug. 2, 2017). The court found Windstream’s review “completely reckless,” noting more than ten percent of the produced documents were privileged and the same documents had been produced twice. The agreement didn’t save the privilege. On reconsideration the following year, the magistrate judge held that the clawback agreement didn’t displace 502(b)(2)’s reasonable-steps duty because the agreement “did not specify what precautionary measures, if any, should be taken to prevent disclosure.” Irth Sols., 2018 WL 575911, at *4 (S.D. Ohio Jan. 26, 2018).

That’s the gap-filler problem. If the order doesn’t expressly tell the court reasonable steps aren’t required, the court will assume they are. The Maxtena court said the same thing more directly: an order or agreement must provide concrete directives regarding each prong of Rule 502(b), or Rule 502(b) will be used to fill in the gaps. Maxtena, Inc. v. Marks, 289 F.R.D. 427, 444 n.16 (D. Md. 2012).

The drafting north star is the Advisory Committee Note’s “irrespective of care” language. An order that incorporates “irrespective of care,” or its functional equivalent “whether inadvertent or otherwise,” signals to the court that the parties have bargained out of the 502(b) reasonableness inquiry. An order without that language leaves you exposed to the Irth Solutions and Maxtenagap-fillers. The court-order shield’s strength depends entirely on what the order actually says.

"Two Cases That Matter"

Leibovic and Captive Alternatives, Side by Side

Two cases, decided six years apart, illustrate the boundary of 502(d) protection. One is what works. The other is what doesn’t.

Leibovic v. United Shore Financial Services. A putative class action over a data breach. On April 24, 2017, XMS produced approximately 1,150 documents from a database compiled by counsel. Two days later, on April 26, XMS notified plaintiff and United Shore that roughly 400 of those documents were non-responsive and were being clawed back as work product under FRE 502 and the parties’ stipulated Discovery Plan, which contained a 15-day clawback notice provision. Judge Roberts enforced the clawback. The 400 inadvertently produced documents, because they were non-responsive and had been compiled by counsel in anticipation of litigation, were protected by the work-product doctrine and subject to clawback under the Discovery Plan. Leibovic v. United Shore Fin. Servs., LLC, No. 15-12639, 2017 WL 3712188, at *3 (E.D. Mich. Aug. 28, 2017). The motion to compel return was denied; the receiving parties had to return the entire production. That’s the cleanest version of how 502(d)/(e) clawback works: agreement in place, notice prompt, documents actually protected.

United States v. Captive Alternatives, LLC. This one is a study in what 502(d) isn’t for. The IRS served an administrative summons on Captive in October 2021, seeking 29 categories of records back to 2011. Captive didn’t comply, and the IRS filed a petition to enforce. During negotiation over roughly 1.1 million potentially responsive documents, Captive moved for a 502(d) order that would have let it produce without privilege review and assert privilege later. Magistrate Judge Tuite denied on three grounds. The summary IRS summons-enforcement proceeding did not involve discovery as that term is understood in civil litigation. The IRS agents investigating Captive were not trained to evaluate attorney-client privilege, and the information needed to make that call was solely within Captive’s possession. The proposed order placed no temporal restriction on when Captive could designate a document privileged, creating an unbounded jurisdictional commitment the court declined to take on. United States v. Captive Alternatives, LLC, No. 8:22-cv-406-TPB-CPT, 2023 WL 5573954, at *3 (M.D. Fla. Aug. 29, 2023).

Read together, the two cases mark the working boundary. A 502(d)/(e) clawback in a Discovery Plan with a clean procedural trigger holds up. A 502(d) order asking a court to bless a no-review production to a requester who can’t evaluate privilege does not. The contrast is the lesson. A 502(d) order is built for adversarial discovery, where each side’s counsel reviews what’s exchanged.

Leibovic v. United Shore (E.D. Mich. 2017)

Posture

Putative class action arising from a data breach.

Mechanism

Stipulated Discovery Plan with a 502(d)/(e) clawback clause and a 15-day notice trigger.

Outcome

Clawback enforced. The court held the 400 inadvertently produced documents protected by the work-product doctrine; the receiving parties returned the entire production.

Takeaway

A clean procedural trigger plus work-product status makes the clawback self-executing.

U.S. v. Captive Alternatives (M.D. Fla. 2023)

Posture

IRS administrative summons enforcement. Not civil discovery.

Mechanism

Motion for a 502(d) order to allow no-review production with later privilege assertion.

Outcome

Motion denied. IRS agents are not trained privilege reviewers, and the proposed order had no temporal cap.

Takeaway

A 502(d) order is not a substitute for privilege review when the requester cannot evaluate the documents.

"Small-Firm Playbook"

Drafting Language That Survives the Meet-and-Confer

Three moves that work for a solo or small-firm federal litigator. None of them is optional.

First, ask for a 502(d) order at the Rule 26(f) conference, not after the first production. Judge Peck’s two-paragraph model order is the cleanest template in circulation. Paragraph one declares that production of privileged or work-product documents, whether inadvertent or otherwise, is not a waiver of privilege in this case or in any other federal or state proceeding, and directs the court to interpret the order to provide the maximum protection allowed by Federal Rule of Evidence 502(d). Paragraph two preserves each party’s right to conduct relevance, responsiveness, and privilege review before production. Two paragraphs. The shorter the order, the harder it is for a court to read a reasonableness inquiry back into it.

Second, write the order to maximize 502(d)’s reach. The phrase “whether inadvertent or otherwise” knocks out the 502(b) gap-filler problem from Irth Solutions and Maxtena. The phrase “any other federal or state proceeding” tracks the statutory portability language. The two clauses together are why the Sedona Conference Appendix A model and the Northern District of California model stipulated protective orders both include them as baseline.

Third, pair the 502(d) order with a documented privilege screening process. The order eliminates 502(b) inside the case. It doesn’t eliminate every collateral challenge, and it won’t help in state-court attacks on the privilege under different rules. The Advisory Committee Note’s recognition of advanced analytical software applications and linguistic tools as potential reasonable steps for 502(b) is your insurance policy.

A 502(d) order is the fallback, not the plan. The plan is a defensible 502(b) record: documented review, sampled QC, and a reviewable audit trail of the calls your team made. Hintyr is an agentic document review platform built for small and mid-size firms. The validation report it generates is the kind of screening record you would want to attach to a 502(b) brief if you ever had to write one. The 502(d) order catches what slips through. The validation record is what keeps the order from being the only thing standing between you and waiver. For the underlying sampling work, see our explainer on technology-assisted review.

"Common Questions"

Frequently Asked Questions

What does a 502(d) order actually protect against?

A 502(d) order protects against privilege waiver when a privileged document is produced in federal litigation. The protection applies in the case where the order is entered, and under the rule’s express language, in any other federal or state proceeding where the same disclosure surfaces.

Do I need both a 502(d) order and a party clawback agreement?

The 502(d) court order is what you need. A 502(e) party agreement that isn’t incorporated into a court order binds only the parties who signed it, which leaves you exposed to third parties, regulators, and future plaintiffs. Most defensible setups put the clawback clause inside a stipulated 502(d) order.

Can I rely on a 502(d) order to skip privilege review entirely?

No. Captive Alternatives shows why courts are skeptical of no-review productions, and Irth Solutions shows why a sloppy review process can sink even a stipulated order. The 502(d) order is a backstop. It isn’t a substitute for documented screening.

What’s the difference between “reasonable steps” and “irrespective of care”?

Rule 502(b) protects you only if you took reasonable steps to prevent the disclosure. A properly drafted 502(d) order can authorize return of the documents irrespective of the care taken, displacing the 502(b) inquiry inside the case. The drafting language matters: orders that don’t expressly disclaim 502(b) often get read as preserving it.

Are state courts bound by a federal 502(d) order?

Under the text of Rule 502(d), a federal court order means the disclosure is also not a waiver in any other federal or state proceeding. Federal courts have given 502(d) orders cross-jurisdictional effect under that language. State-court treatment is less settled, and small-firm litigators planning multi-jurisdiction work should check the analog rule in any state where parallel proceedings are likely.

This article is for general informational purposes only. It does not constitute legal advice and does not create an attorney-client relationship. Statements about Federal Rule of Evidence 502, the Advisory Committee Notes, and the case law cited reflect publicly available sources as of May 2026 and may not address your jurisdiction or matter. Consult qualified counsel before relying on any of the discussion above.

Build belt-and-suspenders privilege defense before the next production.

A 502(d) order on the docket. A privilege workflow your validation report can defend. Hintyr pairs both: agentic document review with the sampling and elusion testing a defensible privilege screen needs. Or compare review platforms side by side before you commit.