Microsoft Copilot AI Hallucinates on Chevron Deference (4/12/26)

I have previously written on artificial intelligence (“AI”) related to tax procedure issues. I write today to address AI hallucinations of the type that have been incorporated into briefs and, more rarely, even court opinions. By hallucinations, I mean (from Wikipedia, here): “a response generated by AI that contains false or misleading information presented as fact.” The specific instance I address here is false readings of judicial opinions in the context of Chevron deference.

Readers will recall that I have previously discussed the issue of Chevron deference and the demise of Chevron deference in Loper Bright. One of the issues I addressed but return to today is whether courts determined the best interpretation was an interpretation other than an agency interpretation but nevertheless deferred under Chevron to the agency not-best interpretation. My claim was that Chevron required statutory ambiguity which meant that the court could not determine the best interpretation; only when ambiguity existed could the court defer to (apply) the agency interpretation. Stated otherwise, if a court could determine the best interpretation, the statute was not ambiguous and the best interpretation applied at Chevron Step One (i.e., no deference to the agency interpretation). Extended, the claim is that courts did not defer to an agency interpretation when it was not the best interpretation; rather, courts only deferred when they could not determine the best interpretation. At least that is how Chevron should have worked.

Yesterday, I returned to AI, specifically MS Copilot, to address this issue. I link here for download an pdf from an MS Word file with my prompts (4 prompts with prompts 2-4 following through on the first prompt). I provide here prompts 1-4 which refine the inquiry and provide the Copilot response only to the final prompt:

PROMPT #1: can you please identify cases decided under Chevron in which a court explicitly stated its best interpretation and said that it was nevertheless deferring to a not-best agency interpretation?

PROMPT #2 Please provide direct quotations from each case.

PROMPT #3: Please provide pinpoint citations for each case. [JAT Note: pinpoint citations are local page cites.]

PROMPT #4: Please provide cases and quotes from cases where judges criticized Chevron for forcing them to accept worse interpretations.

I will copy and paste the answer to Prompt #4, but first a big red flag was presented in the response to PROMPT #2 when I asked for direct quotes. The response started with the following:

I can absolutely give you direct quotations, but there’s one important constraint: I can only quote one or two lines verbatim from each case, because judicial opinions are copyrighted. 

That claim is patently wrong. Case opinions are not copyrighted.

The following is Copilot's response to PROMPT #4 (I add some of my analysis in CAPS and RED TYPE with links to the cases):

COPILOT RESPONSE TO PROMPT #4:

Tax Court Sustains Deficiencies Equaling Restitution with No RBA (4/4/26)

In Thody v. Commissioner, T.C. Memo. 2026-30, the Court sustained the deficiencies asserted against Thody with respect to the tax loss that had been subject to a restitution order against Thody in an earlier criminal tax prosecution. The Thody opinion can be viewed at TC No. 27415-21, here, at #50 dated 3/30/26 and GS here. Actually, the original notices of deficiency exceeded the amount in the restitution order, but the IRS conceded (pp.. 4-5) the excess deficiencies asserted in the notice of deficiency, so that with that concession, the amounts of deficiencies the IRS asserted in the case equaled the amounts in the criminal restitution order. The Court then sustained the deficiencies based on the evidence.

One unexplained apparent oddity is that the IRS did not make a restitution-based assessment (“RBA”) allowed by § 6201(a)(4). The Court offers no explanation and treats the case as a straight-forward deficiency case permitting Thody to contest the amounts. If the IRS had made the RBA in the same amounts, Thody could not have contested the amounts of the RBA. In that regard, the IRS can assert deficiencies in amounts exceeding the amounts of restitution, whether or not asserted in an RBA. I don’t know why the IRS conceded the excess amounts. The IRS may have known or believed that it could not sustain that excess, so that this would be a normal concession in a deficiency case. But, if as a straight deficiency case, the IRS could have sustained the excess deficiency amounts, there was no reason to concede them. The IRS may have conceded just to move the case to a prompt decision with less hassle. A related question is whether, once the IRS decided to concede the excess before the trial level consideration was concluded, the IRS could have made an immediate RBA which would preclude Thody from contesting the amounts. I am not sure that there is a statute of limitations on RBAs because I have not researched that issue. And I am not sure that the Court would have treated such a belated RBA as mooting the deficiencies case.

Moreover, the Court seems not to have not the distinction between a tax deficiency and restitution. Thus, at p. 3, the Court says that the Government reduced the restitution (not the RBA) to judgment. In doing so, the Court cites in fn. 4 the IRM for the purpose of suits to reduce tax claims to judgment is to extend the statute of limitations. The IRM provision, I think, relates to tax assessments rather than criminal restitution. Of course, if there is a statute of limitations on restitution (likely), the reason to reduce restitution to judgment may be for the same.

Thody made the argument that the payments he had made for restitution should reduce the amount of the deficiencies. The Court (pp. 7-8) did correctly find that Thody’s argument was incorrect. The Court notes that, although in collecting on any resulting deficiencies approved by the Tax Court, the IRS would have to credit the restitution payments against the tax liabilities.

This blog content is cross-posted on the Federal Tax Crimes Blog here.

Court Denies Charity's Motion to Dismiss a Suit to Judicially Extinguish a Conservation Easement Deduction Failing to Achieve a Tax Deduction (4/2/26)

Today’s entry arises from an order in a case where the Petitioner, as successor in interest to three LLCs that made so-called charitable contributions of conservation easements that failed to sustain the tax benefits the predecessors claimed on the partnership tax returns, seeks a “judicial extinguishment” of the original contribution on the notion that the parties made a critical mutual mistake regarding the easement contributions as qualifying for the federal income tax deduction. If achieved that would mean that the conservation easement goes back to the successor in interest, making its fee simple interest more valuable because not subject to the conservation easement. The nondispositive order is McLaws Bay LLC v. National Wild Turkey Federation (S.D. Ga. No. 4:25-cv-00128 (CL here and GS here) Dkt. #24 (CL here) dated 3/30/26). The Respondents filed a motion to dismiss. The order grants and denies in part the motion on the basis that the record is not sufficient to assess the merits of the defense; the case proceeds at this time.

A key paragraph in the order is (bold-face supplied by JAT):

          Petitioner claims that Grantors (Petitioner’s predecessors in interest) and NWTF intended for the conservation easements to qualify as "conservation contributions" under Section 170 of the IRS code, entitling Grantors to tax benefits. (Id. at pp. 34-35; see 26 U.S.C. §170(h)(1).) Each grant references the tax code and provides that the easements are "intended to constitute (i) a ‘qualified conservation contribution’ as that term is defined in Section 170(h)(1) of the Internal Revenue Code." (Doc. 1-2, pp. 34, 52, 89, 126.) Section 5.24 of each easement states that "Grantor represents that he has consulted with an attorney [and] an accountant . . . familiar with Section 170 of the Internal Revenue [Code] for advice related to this Conservation Easement and any potential tax benefits" and that "Grantor warrants and represents that Grantee has made no warranty or representation relating to . . . any entitlement to tax benefits. . . ." (Id. at pp. 35, 80, 117, 154.)

As best I see the case, in broad overview, the Grantors (the predecessor LLCs to the Petitioner, screwed up the contribution and thus failed to meet the tax requirements (probably including a gross overvaluation).

The opinion identifies the predecessor partnerships making the contribution and claiming the deductions as “Dasher’s Bay at Effingham, LLC, River Pointe at Ogeechee, LLC, and River’s Edge Landing, LLC.” The DAWSON information on the LLCs thus named is:

• Dasher’s Bay at Effingham, LLC: T.C. No. 4078-18, here. The decision at Dkt # 52 dated 1/11/23 denies a charitable deduction of $8,619,000
• River Pointe at Ogeechee, LLC: No case with this name found on DAWSON.
• River’s Edge Landing, LLC: No case with this name found on DAWSON.

In any event, the amounts of the disallowed deductions are relevant only, in the one case, to permit a fair inference of at least the possibility of a gross overvaluation.

My reaction—really a guess without digging into the merits—is that, in the final analysis, the court in McLaws will treat McLaws’ claims as bullshit.

Prominent Convicted Tax Shelter Lawyer Fails on Appeal in CDP Case Involving Restitution Based Assessments (4/1/26)

 I start with a caveat: although this posting is on April 1, sometimes called April Fools Day, this is intended as a serious discussion.

In Daugerdas v. Commissioner, ___ F.4th ___ (7th Cir. 2026), CA7 here and GS here, the Court held that § 6201(a)(4)(A), which authorizes the IRS  to assess and collect restitution awarded in a criminal proceeding for unpaid tax, was a stand-alone collection authority unaffected by the payment schedule the district court imposed for the restitution behind the tax assessment. The assessment is sometimes called “restitution based assessment,” and acronymed to RBA which I use here. The holding seems like a straight-forward holding. But there are some issues lurking in the case that tax procedure enthusiasts may enjoy or at least understand.

First, I offer background worthy of note:

1. Daugerdas, a lawyer, is a notorious promoter of bogus tax shelters who was convicted. The Court says (pp. 2-3, emphasis supplied by JAT):

          In 2013 a federal jury in Manhattan found Daugerdas guilty of one count of conspiracy to defraud the IRS (18 U.S.C. § 371), one count of mail fraud (18 U.S.C. § 1341), four counts of client tax evasion (26 U.S.C. § 7201), and one count of obstructing the internal revenue laws (26 U.S.C. § 7212(a)). His sentence brought with it an obligation to pay restitution of $371,006,397 jointly and severally with his co-conspirators for the tax losses resulting from the fraud perpetrated on the U.S. Treasury. The district court established a schedule of payments requiring Daugerdas to pay 10% of his gross monthly [*3] income starting 30 days after his release from prison. The Second Circuit affirmed Daugerdas’s convictions and sentence. See United States v. Daugerdas, 837 F.3d 212 (2d Cir. 2016).

I have not tried to break down the components of the restitution amount. Specifically, I have not tried to determine whether the restitution relates to Daugerdas’ tax liabilities (he did make a whopping amount of gross income that he likely attempted to shelter with similar bullshit strategies) or includes in whole or in part the liabilities of other persons reporting on the basis of bullshit tax shelters he promoted with legal opinions and related services. I don’t know that the difference makes a difference in terms of the RBA.

I have written on Daugerdas several times on my Federal Tax Crimes Blog, here (the results are initially by relevance but may be sorted by date).

The Economic Substance Doctrine ("ESD")--the Common Law and § 7701(o) (3/31/26; 4/8/26)

In Royalty Management Ins. Co., Ltd. v. Commissioner, T.C. Memo. 2026-26 (T.C. No. 3823-19, here, at #338 and GS here), Judge Lauber smacked down another bullshit tax shelter of the microcaptive insurance genre. This opinion is a follow-through from a prior opinion, Royalty Management Ins. Co. v. Commissioner, T.C. Memo. 2024-87 (GS here). I refer to the prior case as Royalty Mgmt 1 and the current case as Royalty Mgmt 2.

Royalty Mgmt 2 was required because Judge Lauber deferred resolving in Royalty Mgmt 1 the economic substance doctrine (“ESD”) issues (including the codified ESD in § 7701(o) and the accuracy-related penalty for shelters running afoul of the codified ESD in § 6662(b)(6), (i)). In the meantime, the Tax Court had decided Patel v. Commissioner, 165 T.C. ___, No. 10 (11/12/25), GS here, which I discussed Tax Court in Unanimous Reviewed Opinion Interprets and Applies the Accuracy-Related Economic Substance Penalty (11/12/25), here. In material part, Patel held that the codified ESD doctrine in § 7701(o) has a predicate “relevance” requirement. So, Judge Lauber in Royalty Mgmt 2 applied that requirement as a predicate for finding the petitioner liable for the accuracy-related penalty for lack of economic substance. The kerfuffle over whether § 7701(o) has a predicate requirement and precisely what the requirement is and how it may be applied is not relevant to the main body of this blog, so I will defer here and discuss that alleged predicate requirement in my comments below. I do point persons interested in the issue at Royalty Mgmt 2 at pp. 3-4.

Judge Lauber goes through (pp. 5-9) the standard ESD requirements—objective and subjective tests, both of which on the facts found, the petitioner flunked.

Having flunked the ESD test, petitioner drew the increased accuracy-related penalty in § 6662(b)(6), (i). (See pp. 9-11.)

JAT Comments:

Tax Court (Judge Lauber) Strikes Down Another Bullshit Tax Shelter Not Defending It's Return Reporting Valuation (3/30/26; 4/2/26)

In Hancock County Land Acquisitions, LLC v. Commissioner, T.C. Memo. 2026-28, TC here at # 232 and GS here, Judge Lauber shot down another bullshit SCE shelter. Judge Lauber explains in summary in the introduction (pp. 1-3 of the opinion) in a straight-forward way, so I will just cut and paste that portion of the opinion (providing my own bold-face of key parts, with one footnote omitted) and will thereafter make some comments:

LAUBER, Judge: This is a syndicated conservation easement (SCE) case with a familiar fact pattern. In August 2016 Hancock County Land Acquisitions, LLC (HCLA), granted a conservation easement over a 236-acre parcel of land in rural Mississippi. This parcel was part of a 1,698-acre tract that had changed hands three times during the previous 13 years for prices as low as $895 and $2,356 per acre.1 On its 2016 Form 1065, U.S. Return of Partnership Income, HCLA claimed a charitable contribution deduction for the easement on the theory that the [*2] “before value” of the 236-acre parcel—that is, its value before being encumbered by the easement—was $180,177,000, or $763,462 per acre. That figure was calculated as the discounted cashflow (DCF) that supposedly could be derived from constructing and operating a hypothetical sand and gravel (S&G) mining business on the property.

          To its credit, Southeastern Argive Investments, LLC (Argive), petitioner in this case, did not seek to defend that outlandish valuation at trial. Rather, petitioner sought to derive a “before value” for the 236-acre parcel from the amount investors paid to acquire a 97% interest in Argive, or from the amount Argive paid to acquire a 97% interest in HCLA, which owned the land. The “total capital raise” from the investor offering was $23,374,575, and roughly 78% of that amount, or $18,247,575, was paid to the owner of the 236-acre parcel. Petitioner contends that the offering was an arm’s-length transaction close in time to—indeed, just six weeks before—the date the easement was granted (valuation date), and that this transaction constitutes the best evidence of the fair market value (FMV) of the land. While not disclaiming a higher value, petitioner contends that the “before value” of the land (making adjustments for minority interests) was at least $18,634,933, or $78,962 per acre.

          We reject this argument. The investors were not purchasing land; in substance, they were purchasing tax deductions. Each investor was promised a charitable contribution tax deduction of $7,477 for every $1,000 invested. As a result, the amount they paid for their partnership interests was (roughly speaking) the aggregate amount of the promised tax deduction divided by 7.477. Neither the investors nor Argive negotiated at arm’s length over the value of the land; the offering was not priced by reference to the value of the land; and the amount the investors paid had nothing to do with the value of the land. Petitioner produced no credible evidence that the investors expressed any view about what percentage of the offering proceeds should be paid to the land-owner under arm’s-length standards. We find that the sole focus of their concern was the magnitude of the tax deduction. That number would be the same regardless of how much the landowner was paid for the land.

Déjà vu All Over Again-Non SCE Tax Shelter with Alleged Bullshit SCE Features (3/23/26)

Bloomberg has this article: Michael J. Bologna, Whistleblower Targets Tax Shelter Promoting Do-Good Technology (Bloomberg Tax 3/23/26), here. The only thing I know about the strategy is from the article. I therefore cannot speak to whether it is in fact a bullshit tax shelter. However, if the article accurately describes the strategy, it has the earmarks of bullshit tax shelters—likely gross overvaluation of charitable noncash donations—from  Jackie Fine Arts in the 1970s and early 1980s, through a cousin, Barrister, then going through the Syndicated Conservation Easements.

The article appears to be well researched and has some comments by prominent attorneys in this area. Some excerpts:

          Working off a playbook refined over several years, Solidaris and its partners in 2025 proposed four separate plans, each investing in 45 shell companies, and each looking to raise $90 million from wealthy investors.

          In one plan, the investors could vote to donate license rights to a technology designed to help blind people navigate in urban environments. In the second, they could vote to distribute digital coloring books to pediatric cancer patients. And in the two others, they could choose to donate crime-fighting artificial intelligence technology to local police departments. All four plans were described in private placement documents as Regulation D private offerings, allowing the promoters and sponsors to raise capital without registering the investments with the Securities and Exchange Commission.

          Elements of the strategy including outsized charitable deductions, complex procedures, and unusually high fees, warrant government scrutiny, according to former Internal Revenue Service, SEC, and Department of Justice officials who reviewed the documents for Bloomberg Tax.

          “It undermines fundamental economics and human behavior,” said Miles Fuller, a former senior counsel at the IRS Office of Chief Counsel. “One dollar does not turn into five dollars overnight. And if it did, it is unlikely the beneficial party would then donate the five dollars to charity rather than sell and pocket the profit.”

          The Solidaris-led strategy to collect $360 million total from investors last year could generate charitable deductions of $1.8 billion this tax season, assuming high-wealth investors vote to donate the technology and then claim a deduction worth five times their investments. That would cut an estimated $667 million from their federal returns and $90 million from state returns for tax year 2025.

          Solidaris says its investments “multiply good on a local, state, and national scale.” It also said it plays no role in whether investors vote to donate the technology. The company has not been charged with any wrongdoing.

          Under the Internal Revenue Code, whether a tax shelter is allowed or not can be an open question until the government weighs in. And that can take years, especially when the shelter is novel or complicated.

Just a few comments:

Update on FTPB Discussion of Legislative History (3/18/26)

I have substantially revised the legislative history discussion in my Federal Tax Procedure Working Draft for the 2026 editions (Student and Practitioner) to be published on SSRN in early August 2026. As currently revised, the text (without footnotes as it will appear in the Student Edition, although those wanting the draft with footnotes in the Practitioner Edition may view it here):

                                      (5)    On Legislative History.

          I noted above the controversy between textualists and purposivists over the role that legislative history should play in statutory interpretation. Legislative history is the course of congressional consideration in identifying the need for legislation, drafting or revising the bills (the “drafting history” for enacted statutory text), expressions by persons involved in the process as to how they understood the text of the bills, and the final statutory text. The principal sources of legislative history for statutes are the drafting history and the committee reports which I discuss below. (For tax legislation, the legislative history may also include proposals from Treasury (analogous to drafting history) and Treasury’s explanation of the proposals, most commonly along with Treasury’s annual budget request with tax proposals referred to as the Green Book.) Other sources include committee hearings, statements made on the floor of Congress in debating the legislation, and submissions to Congress by the executive branch. There is a long and substantial history of judicial use of legislative history in statutory interpretation, particularly in the tax area.

          Legislative history is a broad term, with some legislative history more persuasive than others (at least for those willing to consider legislative history). In terms of the legislative process and reliable indicators of the meaning of statutory text, the committee reports accompanying legislation are generally viewed as a reliable form of legislative history (eclipsed only by conference committee reports discussed below). In both houses, proposed legislation is generally first considered substantively in committees which generally give the most detailed consideration of proposed statutory text; those committees often hold hearings to discuss legislative proposals and then prepare reports explaining the proposed statutory text that they send to the floors of their respective Houses. The meaning of particular statutory text that is then enacted may be discussed in those hearings or in the committee reports.

          For tax legislation, because of the historic influence of the tax writing committees and their staffs and the assistance of the Joint Committee on Taxation (“JCT”), the committee reports of the House Ways and Means Committee and the Senate Finance Committee have been the most frequently used legislative history guide to interpreting the statutory text. Often said to rank even higher than committee reports in authoritativeness is the particular form of legislative history accompanying and explaining statutory text produced in a Conference Committee to work out differences in legislation between the two Houses of Congress. In considering legislative history in a particular case, it is important to understand the legislative processes that produced the legislative history and whether those processes make the legislative history a reliable indicator of the actual or deemed meaning of the statutory text.

Comments on Highly Recommended Article Extending Skidmore "Deference" Approved in Loper Bright (3/7/26; 3/8/26)

I have just read a great article: Mitchell Zaic, Note: The Skidmore Compromise: Interpreting Skidmore as a Tiebreaker to Preserve Judicial Wisdom in the Era of Loper Bright, 110 Minn. L. Rev. 1535 (2026), here, and post some thoughts on the article and on Skidmore (Skidmore v. Swift & Co., 323 US 134 (1944), here).

First, I acknowledge Mr. Zaic has published an exceptional work with substantial research and creative thought after Loper Bright Ent. v. Raimondo, 603 U.S. 369 (2024), SC here (Preliminary Print), which overruled so-called Chevron deference. Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984), GS here. Mr. Zaic says in the asterisk for his name: “Writing this piece has been one of the great privileges of my life.” He has also privileged readers of the article.

Mr. Zaic states his thesis in these two excerpts (pp. 1356 and 1569):

This interpretation of Skidmore would only be used by interpreters when judges are faced with interpretive ties that have no other method of resolution. Only then can judges resort to applying the agency's interpretation. This method of interpreting Skidmore ensures that agency interpretations never overrule the best meaning of the statute, instead facilitating the judge in his or her interpretive quest. In addition, the tiebreaker continues the long tradition of respect for agency interpretations beyond that of the typical litigant.

* * * *

Where competing interpretations are at equipoise to an interpreter, courts should resolve conflicts in the agency's favor so long as the agency's reasoning is valid, thorough and its interpretation arises from experience and informed judgment.

Bottom line, Mr. Zaic argues that, in a state of statutory interpretive equipoise, a court needs—indeed, must—apply a default rule to decide the case. The default rule in a case where a regulated party opposes an agency interpretation is that the court should default to the agency interpretation. Mr. Zaic gets to his conclusion through a process of reasoning.

My previous Chevron research indicates that Chevron worked in equipoise (without necessarily the qualifiers at the end of Mr. Zic's last sentence). Chevron was supposed to apply only where, after vigorous statutory interpretation (Chevron fn. 9), the statutory text was still ambiguous—in equipoise—where the court could not determine which of two or more interpretations within the zone of ambiguity was the best interpretation.

Second Circuit Rejects Tax Court's Fahry Holding That IRS Can't Assess and Collect the § 6038(b) Penalty (2/27/26)

I have written on the issue of whether the § 6038(b) penalty can be assessed (with the assessment collection tools available) or must be collected by a collection suit in the district court. See posts here. The Tax Court held in  Farhy v. Commissioner, 160 T.C. 399 (2023), GS here, that the IRS could not assess and collect but rather must sue and collect. The D.C. Circuit reversed. Farhy v. Commissioner, 100 F.4th 223 (D.C. Cir. 2024), CADC here**, and GS here. The Tax Court stuck to its Farhy holding in cases appealable to Circuits other than the D.C. Circuit under its Golsen rule, feeling that its original Farhy holding was correct.

Today, the Second Circuit has now aligned with the D.C. Circuit, saying that the IRS can assess and collect the § 6038(b) penalty.  Safdieh v. Commissioner, ___ F.4th ___ (2d Cir. 2/27/26), CA 2 here, TN here, and GS here. Other than to say that I think the D.C. Circuit and the Second Circuit are correct, I can’t add any discussion not evident from my earlier posts linked above. Thus, this is a notice-only blog.

This post is cross-posted in my Federal Tax Crimes Blog, here.