Frontier challenges in AI tax prep: Trust & estate K-1s

In the first installment of this series, we walked through the frontier challenges of K-1 processing, focused on partnership and S corporations, which account for the bulk of K-1 volume at most firms. This post is about their lower-volume, higher-risk counterpart: the Schedule K-1 from Form 1041, issued by trusts and estates.

Trust and estate K-1s are lower volume. They're also, box for box, harder to route correctly on the 1040. The 1041 K-1 has 14 boxes compared to 21 on the 1065, but those boxes carry items that route to different schedules, interact with different phase-outs, and follow different rules than their partnership equivalents. The 1040 preparer takes the K-1 at face value, as they should. The challenge is knowing where each item goes, what else it affects downstream, and which boxes require treatment that breaks from the partnership K-1 muscle memory most preparers rely on.

To make this concrete, we'll follow a single trust through the entire post: The Whitfield Family Trust, an irrevocable non-grantor trust created in New York, administered by a California-based corporate trustee, with two beneficiaries. Sarah receives mandatory income distributions (Tier 1). James receives discretionary distributions (Tier 2). The trust holds rental real estate, a dividend portfolio, and municipal bonds. Last year's K-1 to Sarah showed $45,000 of ordinary income and $12,000 of qualified dividends.

This year, Sarah's K-1 looks different. That difference is where trust in K-1 preparation gets harder.

Why Trust K-1s Are Different

Trust K-1s are one example of a recurring pattern in tax preparation: specialized document types that require specialized routing logic, specialized state rules, and specialized review surfaces. Firms often handle these cases through a mix of tax software inputs, preparer judgment, reviewer notes, and manual follow-up. Accrual is built to bring that work into one preparation flow: read the document, route the values, preserve the source support, and show the reviewer what needs attention. The goal is not a separate tool for every special case. It is one system that prepares the return from the source documents, routes each item to the right field, keeps the supporting document visible, and shows reviewers what changed.

The 1040 preparer's job with any K-1 is the same: take the numbers at face value and route them to the right places on the return. With a partnership K-1, the routing is familiar. Most preparers handle 1065 K-1s regularly enough that the muscle memory is reliable. Trust K-1s break that muscle memory in three ways.

First, the routing rules are different box by box. Items that look similar to partnership K-1 items (capital gains, ordinary income, deductions) go to different places or interact differently with the rest of the 1040. Box 10 has no partnership equivalent. Box 11's sub-categories changed after TCJA and each routes to a different line. Box 14a is non-taxable but affects downstream computations that change the return's bottom line. A preparer defaulting to partnership K-1 habits will place trust K-1 items in the wrong fields, and the errors are subtle enough to survive review. The K-1 may be entered. The issue is whether it was prepared correctly.

Second, trust K-1 amounts are more volatile year over year than partnership allocations. The income on Sarah's K-1 from the Whitfield Trust is a function of the trustee's distribution decisions, the trust's income mix, and the "distributable net income", or "DNI" computation, all of which can shift significantly between years. A 60% drop in ordinary income that would be a red flag on a partnership K-1 might be perfectly normal for a trust where the trustee exercised discretion differently. SALY is a weaker heuristic for trusts, and a preparation system that doesn't calibrate its reasonableness checks for trust-level volatility will produce false positives constantly. The system needs to know when a year-over-year change is normal, and when it is worth putting in front of a reviewer.

Third, the system has to classify the trust correctly before routing anything. Grantor trusts bypass the standard 1041 K-1 framework entirely. All income is taxed to the grantor under §§671-679, and the reporting comes in at least three formats: a K-1 from an informational 1041, a grantor trust letter with no standardized layout, or income reported directly on the grantor's SSN with no 1041 filing. Partially grantor trusts split income across both treatments. An AI system that doesn't identify the trust type before it starts routing will apply the wrong framework to every box on the K-1.

Why the Numbers Can Change Between Years

The 1040 preparer doesn't need to re-derive the DNI computation. But they do need to understand why the character mix on a trust K-1 can shift between years even when the trust's underlying investments haven't changed. On a partnership K-1, if the entity earned qualified dividends, the partner reports qualified dividends. The relationship is direct. On a trust K-1, the character is the output of a proportionate DNI allocation. If the trust's income mix shifts, or the trustee changes the distribution amount, the character on every beneficiary's K-1 shifts with it.

For the Whitfield Trust, this means Sarah's K-1 might show $12,000 of qualified dividends one year and $24,000 the next, not because the trust doubled its dividend income, but because the trustee increased her distribution and the proportionate share of each income type grew with it. A preparer who flags that jump as suspicious is wasting review time. A preparer who doesn't flag it when the underlying cause is actually a problem (a miscoded K-1, a changed beneficiary classification) misses something real. Since James receives only discretionary income, he might not even receive a K-1 next year.

The practical implication for AI tax preparation is that reasonableness checks on trust K-1s need wider tolerances than partnership K-1s, and the system needs to track the prior-year baseline per trust, per beneficiary, so it can distinguish normal trust volatility from genuine anomalies worth a preparer's attention. The point is not to make reviewers chase every difference. It is to show them the differences that matter.

Separately Stated Items: Where the Routing Gets Tricky

The 1041 K-1's boxes carry items that need to be routed to different schedules and lines on the 1040, and the routing logic differs meaningfully from partnership K-1s. Each of the following has tripped up experienced preparers we've talked to. The issue is not whether the system can read the box. The issue is whether it knows what that box does to the return.

Capital gains (Boxes 3 and 4). Capital gains on trust K-1s appear and disappear between years. Whether the trust distributes capital gains depends on the governing instrument and trustee decisions, and that can change year to year. Suppose the Whitfield Trust's K-1 to Sarah shows $15,000 of long-term capital gain in Box 4a this year for the first time. That routes to Schedule D on her 1040. The preparation challenge is what happens next year. If the trustee reverts to retaining gains at the trust level, Box 4a may go blank. A preparer might waste time chasing the "missing" capital gains or, worse, carry a number forward that doesn't belong. This kind of year-over-year variability is normal for trusts and unusual for partnerships, so the system needs to handle both the presence or absence of capital gains on a trust K-1 without treating either as an error.

Estate tax deduction (Box 10). This is the item preparers miss most often. When a beneficiary receives income in respect of a decedent (IRD), the estate may have paid estate tax on the same income. Box 10 provides a deduction to prevent double taxation. Individual beneficiaries generally claim it as an itemized deduction on Schedule A, line 16, with special computation effects when the IRD is capital gain. The deduction applies when an estate or trust distributes IRD and there was actual estate tax attributable to those IRD items. We've heard from reviewers that Box 10 gets skipped because it appears infrequently and preparers don't encounter it often enough to build reliable habits around it. Missing it overstates the beneficiary's tax liability, sometimes by thousands of dollars.

Tax-exempt interest (Box 14a). The Whitfield Trust's municipal bond interest flows through to Sarah and remains tax-exempt, but it still matters. It affects the taxability of Social Security benefits, increases MAGI for various phase-outs, and must be reported on the return even though it isn't taxed. Preparers who skip Box 14a because the income is "non-taxable" produce a return that calculates correctly on its face but gets the downstream computations wrong. It's a subtle error that only surfaces when the IRS recomputes Social Security taxability or a phase-out threshold.

Excess deductions on termination (Box 11). When a trust or estate terminates, unused deductions may pass to the beneficiaries. Post-TCJA, the rules changed: only Section 67(e) expenses and non-miscellaneous deductions are allowable. Box 11 now breaks this into sub-categories: 11a for Section 67(e) excess deductions, 11b for non-miscellaneous, and 11c through 11f for capital loss and NOL carryovers. A preparer working from pre-2018 habits will place the full amount as miscellaneous itemized deductions, which is no longer correct. The sub-categories each route to different places on the 1040, and the distinction matters on audit.

Final K-1s and One-Time Items

The "Final K-1" checkbox is the single most important signal on a 1041 K-1 for routing purposes, and it's where preparers most often stumble.

If the Whitfield Trust were to terminate, Sarah's final K-1 might include capital loss carryovers in Box 11c/d and net operating loss carryovers in Box 11e/f that pass to her as if they were her own. These are one-time items that route to specific places on Sarah's 1040 and should never appear again. A preparer who carries them forward to the following year's return will either double-count them or, more commonly, wonder why next year's K-1 doesn't include them and waste time investigating. A preparer who misses the "Final" checkbox entirely may not apply the termination routing rules at all, treating the excess deductions and carryovers as ordinary recurring items.

A prior-year baseline is what tells normal volatility from a real anomaly

Accrual's approach uses prior-year return data to establish a baseline for each trust K-1, while the planning layer flags the situations that warrant preparer attention: a new trust appearing for the first time, a familiar trust with dramatically different distribution amounts, or a final K-1 with termination items that require one-time routing.

State Complexity for Trusts

State taxation of trusts adds another layer that plays out clearly in the Whitfield Trust scenario. Unlike individuals, where residency is determined by physical presence and domicile, trust residency varies by state across at least four different nexus theories:

• Where the trust was created. New York taxes trusts created by New York domiciliaries, regardless of where the trust is now administered.
• Where the trustee resides. California looks to trustee residence as a primary nexus factor.
• Where the beneficiary resides. Some states tax trust income to the extent it's distributed to in-state beneficiaries.
• Where the trust is administered. A minority of states focus on the situs of administration.

The Whitfield Trust was created in New York with a California-based corporate trustee. Sarah lives in New Jersey. James lives in Texas. The trust may owe tax to states Sarah has no personal connection to, and her K-1 may show withholding from those states. Whether that withholding creates a credit on Sarah's New Jersey return depends on New Jersey's rules for trust income, not on what the K-1 says.

For the 1040 preparer, the point is simple: state amounts on a 1041 K-1 cannot be treated like state amounts on a 1065 K-1. The state listed on the K-1 reflects the trust's tax obligations, not necessarily the beneficiary's resident state treatment.

Accrual supports state-specific trust K-1 handling, including Illinois Schedule K-1-T, and applies the same three-layer architecture to state routing: understanding the state supplement, planning the routing based on the beneficiary's residency, and populating the correct state fields in the return.

What Preparers Get Wrong

To make the stakes concrete, here are two errors we've seen experienced preparers make. In both cases, the K-1 was read correctly. The mistake was in how the return was prepared. That distinction matters. Trust and estate K-1 risk often sits after extraction, in the routing, treatment, and review of the return.

The Box 10 deduction that nobody claimed. An estate K-1 reported a Box 10 deduction of $8,200 for the estate tax attributable to IRD. The preparer, handling dozens of K-1s during peak season, entered the income items from the K-1 but skipped Box 10. It's an unusual item that doesn't appear on most K-1s, and the preparer hadn't encountered one since the prior year. The beneficiary was taxed on IRD items that had already been subject to estate tax, overstating their liability by roughly $3,000. A reviewer scanning the K-1 input saw the income items handled correctly and approved the return. The deduction was only caught when the client's attorney asked about it months later.

Excess deductions routed to the wrong line. A trust's final K-1 reported $14,500 in excess deductions: $9,200 of Section 67(e) expenses in Box 11a and $5,300 of non-miscellaneous deductions in Box 11b. The preparer, who had been handling trust K-1s since before TCJA, placed the full $14,500 as miscellaneous itemized deductions, following the pre-2018 treatment. Under current rules, the Section 67(e) expenses should have been reported as an adjustment to income, and the non-miscellaneous deductions as separate Schedule A line items. The return's total tax happened to come out close to correct, which made the error invisible in review. It would have surfaced on audit.

Both errors are understandable. Box 10 appears infrequently enough that even experienced preparers don't build consistent habits around it. The Box 11 sub-category rules changed in 2018 and the old muscle memory persists. These are exactly the kind of errors that AI preparation systems should catch: low-frequency, high-consequence items where human pattern recognition breaks down because the pattern doesn't repeat often enough to become reliable.

Questions Worth Asking When Evaluating Tools

If your firm processes trust and estate K-1s and you're evaluating AI preparation tools, these questions cut to the capabilities that matter:

Can it handle excess deductions on termination under current post-TCJA rules? The Box 11 sub-categories (Section 67(e), non-miscellaneous, capital loss carryovers, NOL carryovers) each route to different places on the 1040. This changed in 2018, and many systems haven't caught up.

Does it recognize and properly route the estate tax deduction for IRD? Box 10 is one of the most commonly missed items on trust and estate K-1s. If the tool can't handle it, preparers are doing manual work on every estate K-1.

Can it flag when a "Final K-1" requires special handling? Termination items are one-time events that change the routing rules for that K-1. The system should recognize the flag and adjust.

Does it handle state trust taxation? Trust residency rules are state-specific and fundamentally different from individual residency. If the tool assumes the trust's state is the beneficiary's state, every multi-state trust K-1 is a potential error.

Can it show the reviewer where each value came from, what changed from the prior year, and which items need judgment? That is a different question than whether the tool can extract the K-1. Extraction is necessary. It is not sufficient.

Where This Is Headed

Trust and estate K-1s will never be the highest-volume item in a firm's workflow. But they are, per unit, among the most error-prone. Separately stated items with unique routing rules, post-TCJA changes to excess deductions, the estate tax deduction that experienced preparers still miss, and state nexus rules that follow entirely different logic from individual residency make every 1041 K-1 a small puzzle that partnership K-1 habits don't solve. That's the platform thesis we laid out at the top of this post: one preparation architecture that handles every K-1 type correctly, rather than requiring a different manual process for each one.

Today, Accrual prepares the 1040 side: routing trust K-1 items to the right fields, flagging anomalies against the prior-year baseline, and catching the low-frequency items that human muscle memory misses. But for firms that prepare both the 1041 and the beneficiary's 1040, there's a natural next step. When the same platform handles both sides, the K-1 isn't just a document to route. It's an output the system can verify against the source computation. That's where we're headed.

Not a separate workflow for every case. One system that prepares the return, keeps the source support visible, and shows reviewers the items that need judgment.

If you're dealing with complex trust and estate K-1 workflows and want to see how Accrual handles the cases that require more than extraction, reach out to our team or contact us.

This post is part of Accrual's CPA Series, where we explore the technical accounting challenges behind AI-powered tax preparation.