On 20 July 2026 the Upper Tribunal released (1) Opus Labour Services Limited (in liquidation) (2) Jason Giller v HMRC [2026] UKUT 00275 (TCC). Judges Thomas Scott and Andrew Scott dismissed every ground of appeal. The company keeps the Kittel denial of input VAT. The sole director keeps the Personal Liability Notices.
If you buy labour, outsource payroll, use umbrellas, run CIS chains, or sign off supplier due diligence as a once-a-year pack, this is not a niche recruitment case. HMRC can strip input VAT, layer penalties, and put the personal bill on the director who “should have known” — even where the tribunal stops short of calling him a fraudster.
The GOV.UK decision page is here. The underlying FTT decision is Opus Labour Services Ltd & Anor v HMRC [2025] UKFTT 800 (TC). Practitioner notes are already out from Claritax and ContractorCalculator. What follows is the CFO cut.
What Actually Happened
Opus was a temporary employment agency supplying construction labour, including asbestos removal crews. It bought payroll services from a series of specialist providers. Those providers paid the workers, invoiced Opus with VAT, and Opus reclaimed the input tax in the ordinary way.
HMRC said the payroll companies were defaulting traders, or buffers in chains that led to a VAT loss. Before the FTT, tax loss, fraudulent evasion and connection were conceded. The only live issue was knowledge: did Opus, through Mr Giller, know or should it have known that the transactions were connected with fraudulent VAT evasion?
The FTT’s answer was careful and damaging:
- Actual knowledge: no. The tribunal would not brand Mr Giller a knowing fraudster over an extended period while under HMRC scrutiny.
- Blind-eye knowledge: no, “with some reservations.” It did not find he suspected facts and deliberately refused to confirm them.
- Constructive knowledge: yes. He should have known. Failure to make proper checks was “seriously negligent.” Five consecutive fraudulent counterparties was not bad luck.
HMRC denied input tax, imposed company penalties, and issued two Personal Liability Notices under the VATA regime. Reporting around the case puts personal exposure above £900,000. Company insolvency does not automatically end the story for the person who ran the checks.
The Legal Test the UT Just Re-Anchored
The right to deduct is lost where a taxable person “knew or should have known” that, by the purchase, they were taking part in a transaction connected with fraudulent VAT evasion. That is the CJEU principle in Kittel (Joined Cases C-439/04 and C-440/04), as applied in the UK through the Court of Appeal in Mobilx Ltd v HMRC [2010] EWCA Civ 517.
HMRC’s manual at VATF53420 still quotes Moses LJ’s “only reasonable explanation” language. Taxpayers have spent years treating that phrase as a separate, higher hurdle HMRC must always clear in a particular verbal form.
The Upper Tribunal is blunt that this over-refines the test. The question remains simple: knew or should have known. “Only reasonable explanation” is one way to prove constructive knowledge, not a mandatory magic formula, and HMRC need not invent every alternative commercial story the taxpayer never ran. Later UT authorities such as GSM Export and AC Wholesale Ltd v HMRC [2017] UKUT 191 (TCC) already pointed that way.
Grounds 1 and 2 — wrong test / inadequate reasons — failed. Ground 3 — Edwards v Bairstow challenges on prior industry awareness, thin supplier checks and the repeated pattern of failures — also failed. The UT refused late new evidence from a former Ebrit director under Ladd v Marshall. You do not get a second evidentiary bite on appeal because the FTT strategy underperformed.
Why This Hits Broader Than Recruitment Agencies
Labour supply is the fact pattern. The control failure is general. The FTT’s findings, left undisturbed, are a due-diligence autopsy:
- prior awareness of supply-chain integrity risk from earlier industry experience;
- inadequate commercial checks before and during relationships with payroll providers;
- a repeating pattern of supplier failure that a reasonable business person should have treated as a red flag, not noise;
- an attitude that due diligence was “guidance” rather than the price of keeping the company’s money and the director’s balance sheet intact.
If your model depends on third-party payroll, umbrellas, CIS chains or high-velocity labour suppliers in a fraud-sensitive sector, Opus shows how “we paid the invoice and reclaimed the VAT” becomes “you funded someone else’s default.”
It sits next to the wider 2026 labour-supply push. From April 2026, agencies face sharper exposure where umbrella PAYE and NIC are not properly accounted for. Those rules are not a Kittel rewrite. They do not need to be. VAT can already create parallel corporate and personal ruin on the same architecture. Due diligence that tests PAYE/NIC and ignores VAT settlement is half a control framework.
What “Should Have Known” Looks Like in a Finance File
Tribunals do not need a smoking-gun email. They build an objective picture from circumstances. In Opus, the damning combination was sector risk + thin checks + serial defaults + weak answers under cross-examination.
- Onboarding that is commercial, not theatrical. Companies House extracts and a VAT number screenshot are not a risk assessment. Credit, ownership, banking behaviour, capacity, references, and whether the economics make sense without retained VAT all matter.
- Ongoing monitoring, not a one-off pack. Kittel is assessed against the circumstances of the transactions. A clean kick-off memo from 18 months earlier does not immunise you after three failed counterparties in the same role.
- Pattern recognition as a control. One bad supplier can be misfortune. A conveyor belt of payroll providers that disappear, default or get replaced under pressure is evidence. Treat it that way in the board minutes.
- No outsourced conscience. “My accountant didn’t tell me” and “HMRC didn’t prescribe the checklist” cut no ice in the FTT reasoning. Commercial common sense is part of the test.
- Document the no as well as the yes. When you walk away because terms are too soft, ownership is opaque, or VAT behaviour cannot be evidenced, keep the note. The absence of walk-aways is itself a fact pattern.
HMRC’s fraud manuals make the same point from the other side of the table: general awareness, contrived features, ignored negative indicators, and “too good to be true” economics all feed the constructive-knowledge analysis. See VATF53420 and the linked material in the VAT Fraud manual.
Personal Liability Is the Real Escalation
Company-level Kittel denials are painful. Director-level PLNs change careers. Where penalties are imposed and HMRC can attribute the company’s failure to a director’s conduct, personal liability notices under VATA can make the individual pay. In a liquidation, that is often the only solvent pocket left. Opus is a clean illustration: company in liquidation, director still fighting, UT still dismissing.
For PE-backed portfolio companies and founder-led agencies alike, revisit delegated authority for supplier approval, D&O and tax insurance assumptions, exit warranties where labour supply is material, and any structure that assumes the operating company contains the risk. If the file cannot show a reasonable director would have stopped, re-checked or exited, “the company took the risk” is not a complete answer.
CFO Actions This Month
- Map every labour and payroll counterparty. Direct agencies, umbrellas, payroll bureaux, CIS chains, preferred-supplier lists. Identify who invoices VAT to you and who actually pays the workers.
- Rebuild due diligence as a living control. Onboarding checklist, periodic refresh triggers, adverse-media and insolvency watches, and a hard stop when a provider cannot evidence timely VAT accounting.
- Stress-test commercial terms. Extended unsecured credit from a payroll provider is not automatically fraudulent. It is a risk signal. Ask how working capital is really funded.
- Quantify Kittel / PLN exposure. Input VAT reclaimed on high-risk categories over open periods; penalty ranges; which individuals signed off the relationships; what the board papers actually said.
- Align VAT, PAYE, NIC and CIS reviews. Do not run three separate comfort memos on one supply chain. One operating model, one risk file.
- Brief the board in plain English. In Opus, “should have known” was serious negligence against objective red flags, not a finding of deliberate fraud. Boards understand that language.
- If HMRC is already writing, freeze the narrative. No casual emails, no retrospective diligence theatre, no new witness strategy invented after the FTT has found the facts. Get specialist VAT litigation support early.
Bottom Line
Opus does not invent Kittel. It applies it to a modern labour-supply model and refuses to let appellate wordplay unwind a fact-heavy FTT decision. Constructive knowledge does not need a particular Mobilx catchphrase. Five failed counterparties plus thin checks was enough.
If your margin depends on someone else running payroll cleanly, your controls have to prove you cared whether they did. Otherwise HMRC can take the input VAT, the penalties, and — through PLNs — a slice of the director.
If you want a structured review of labour-supply VAT exposure, supplier due diligence, or director personal-liability risk across a portfolio company, get in touch.