LZLabs v IBM: Contractual Bars Trump Software-Directive Defences – A High Hurdle for Fact-Heavy TCC Appeals

1. Introduction

LZLabs GmbH & Others v IBM UK Ltd ([2025] EWCA Civ 842) is the Court of Appeal’s refusal to grant permission to appeal (“PTA”) from an extensive Technology & Construction Court judgment ([2025] EWHC 532 (TCC)) delivered by O’Farrell J. The dispute arose from LZLabs’ attempt to build a “Software Defined Mainframe” (SDM) capable of running IBM mainframe applications without IBM hardware. Using a small English affiliate (Winsopia) LZLabs licensed a suite of IBM programs under a standard IBM International Computing Agreement (“ICA”) – a licence which, in explicit terms, forbade reverse assembly, decompilation or other reverse engineering.

At first instance IBM succeeded on every material issue: numerous breaches of the ICA, unlawful procurement of those breaches and unlawful means conspiracy. Comprehensive injunctions, delivery-up/destruction orders and a £20 million interim costs order followed. The appellants lodged a sprawling “kitchen-sink” draft appeal – 35 headline grounds blossoming into some 150 sub-grounds – and sought urgent expedition because the destruction date loomed.

The Court of Appeal (Coulson LJ and Males LJ concurring) refused PTA in its entirety. Although technically a PTA decision, the court produced a lengthy, reasoned judgment which clarifies important points of substantive intellectual-property law, contract construction and appellate practice. Two themes dominate:

1. Substance – Contractual prohibitions on reverse engineering survive the EU Software Directive’s “observe-study-test”, “interoperability” and “error-correction” exceptions; Article 5(3) or 6(1) cannot be repurposed to legitimise wholesale disassembly.
2. Procedure – In fact-dense, expert-heavy TCC litigation the appellate hurdle is “particularly high”; attempts to re-run findings of fact and evaluation under a barrage of grounds will not be entertained.

2. Summary of the Judgment

• Main findings re-affirmed: The appellants systematically breached explicit ICA clauses and deliberately mined IBM code to accelerate SDM development; LZLabs and its owner Mr Moores knowingly procured those breaches and participated in an unlawful means conspiracy.
• Software Directive defences rejected: Article 5(3) (observe/study/test), Article 6 (interoperability decompilation) and Article 5(1) (error correction) did not displace the ICA bar. “Reverse assemble/compile/engineer” is not the same as “observe, study or test”.
• High PTA threshold emphasised: Five converging considerations – factual findings, evaluative decisions, expert evidence, specialist TCC context and breadth of the proposed appeal – made permission “particularly” difficult to obtain.
• No ‘other compelling reason’: The case turned on its facts; no broad public-interest issue arose.
• Injunctions upheld (with a minor drafting tweak): The SDM and any information “derived from breaches as found by the judge” remain enjoined; ICA program copies to be delivered up or destroyed.

3. Analysis

3.1 Precedents Cited and Their Influence

1. Fage v Chobani [2014] EWCA Civ 5 – Re-affirmed the limited appellate review of primary findings; Coulson LJ quotes Lewison LJ’s “island-hopping” metaphor to stress why a 33-day trial’s factual matrix could not sensibly be reopened.
2. Re Sprintroom [2019] EWCA Civ 932 – Sets the test for challenging evaluative findings; must show an “identifiable flaw” such as a gap in logic. Used to dismiss grounds alleging the judge’s evaluative conclusions were “irrational”.
3. Thompson v Christie’s [2005] EWCA Civ 555 – Warns appellate courts against “cherry-picking” expert points within a fact/expert jigsaw. Anchors reluctance to reconsider O’Farrell J’s interwoven technical findings.
4. Wheeldon v Millennium Insurance [2018] EWCA Civ 2403 – Coulson LJ’s own synthesis of appellate restraint in TCC cases; forms part of the “particularly high permission hurdle”.
5. Lifestyle Equities (SC, 2023) – Clarifies mens rea for accessory liability: knowledge of essential facts, not the legal characterisation, suffices. Deployed to defeat arguments that a mistaken view of the Software Directive negated intent.
6. InfoPaq (CJEU, 2009) – Principle that copyright exceptions are construed narrowly. Approved to rebut the appellants’ expansive reading of Articles 5 & 6.
7. Perry v Raleys [2019] UKSC 5 – Quoted for the limited bases on which appellate courts may disturb primary findings: no evidence or wholly unreasonable findings.

3.2 Court’s Legal Reasoning

3.2.1 Contract v Directive

O’Farrell J accepted that the ICA must be construed against the Software Directive to avoid illegality, but emphasised that the Directive’s mandatory derogations are limited. The Court of Appeal endorsed her two-step approach:

1. Identify whether the acts complained of breach the express licence terms.
2. Consider whether the Directive requires those terms to be disregarded for the specific acts.

Reverse engineering undertaken to extract implementation detail does not fall within Article 5(3)’s observe/study/test right, which contemplates passive study of functionality while in authorised use. Likewise, Article 6 decompilation is only permitted insofar as it is “indispensable” to achieve interoperability of an independently created program – a purpose not evidenced here. Because the disassembly served SDM development (a competing product), it was “extreme” and unprotected.

3.2.2 Construction of Key ICA Clauses

• “ICA Program” unequivocally covers “the original and all whole or partial copies”. No “substantiality” threshold exists; importing one would contradict the express words.
• Clause 4.1.3(a) prohibits “reverse assemble, reverse compile, otherwise translate or reverse engineer” – terms interpreted with expert assistance; the court rejected an attempt to limit breach to literal copying.
• Clause 4.1.3(b) (“transfer… outside Winsopia’s enterprise”) extends to any component, however small, because the definition already includes “partial copies”.
• Audit & Termination – IBM’s audit request was reasonable; Winsopia’s confidentiality objection unsustainable because much requested material was not secret and alternative safeguards existed.

3.2.3 Accessory Liability Principles

Applying Lifestyle Equities, the court held that LZLabs and Mr Moores needed only knowledge of the facts constituting Winsopia’s breach (use of IBM programs in prohibited ways) and intention to induce those acts; belief about legal consequences was irrelevant. Documentary evidence showed active steps to dilute “Clean Room” safeguards – sufficient to prove procurement and conspiracy.

3.2.4 Appellate Practice – The ‘Particularly High Permission Hurdle’

Coulson LJ synthesised five overlapping factors raising the PTA bar:

1. Challenge predominantly to primary facts.
2. Where findings are evaluative, threshold still high.
3. Heavy reliance on expert evidence.
4. Decision from a specialist court (TCC).
5. “Kitchen-sink” appeal weaponising 150 sub-grounds.

Combined, these make re-litigation impractical and unjustified.

3.3 Potential Impact

• Software Licensing – Confirms that UK courts will enforce detailed prohibitions on reverse engineering even post-Directive. Licensors may draft explicit clauses with confidence that Article 5 or 6 cannot be exploited to justify program disassembly.
• Clean-room Projects – Provides judicial guidance that superficial “containment” procedures will be scrutinised; watering them down risks accessory liability for company principals.
• Appeals from the TCC – The judgment will be cited to discourage unfocussed, fact-heavy appeals; advocates are placed on notice that prolix grounds and overlong skeletons may backfire.
• Injunction Practice – Affirms the court’s willingness to order destruction/delivery-up of infringing software and to enjoin derived information, extending beyond literal code copying.

4. Complex Concepts Simplified

EU Software Directive (Directive 2009/24/EC)

Provides copyright-style protection for computer programs, but contains exceptions:

• Article 5(1): User may make necessary copies for “correcting errors”.
• Article 5(3): User may observe/study/test functioning whilst lawfully using the program.
• Article 6: Permits limited decompilation when indispensable to obtain interface information for interoperability with an independently created program.

The Directive states contractual terms contrary to these exceptions are “null and void”.

Reverse Engineering vs Observe/Study/Test

“Observe, study or test” covers passive analysis while running the software. Reverse engineering, decompilation or disassembly require active conversion of object code to human-readable form – a step beyond mere observation and therefore usually outside Article 5(3).

Unlawful Procurement of Breach

A party that persuades, induces or facilitates another to break a contract is liable if (i) it knew the essential facts rendering the act a breach and (ii) intended to procure it. Knowledge of legal wrongfulness is unnecessary.

Unlawful Means Conspiracy

A combination of two or more persons who engage in unlawful means (here, contractual breaches) intending to harm the claimant. At least one conspirator must commit the unlawful act; others are liable if they participate with requisite intent.

Permission to Appeal (Civil)

Granted only where the appeal has a “real prospect of success” or there is “some other compelling reason”. Factual findings are disturbed only if Perry v Raleys criteria met: no evidence, or plainly unreasonable conclusion.

5. Conclusion

The Court of Appeal’s refusal of PTA in LZLabs v IBM is more than a procedural footnote. It restates, with modern software-industry flavour, that:

• Licence terms prohibiting reverse engineering will be enforced unless the alleged acts fall squarely within narrow Directive exceptions – and they rarely will if the aim is competitive product development.
• Company principals cannot shield themselves behind purported misunderstandings of EU directives; knowledge of facts, not of law, suffices for accessory liability.
• Fact-heavy, expert-driven TCC judgments enjoy strong appellate deference. Parties should shape potential appeals around genuine questions of law, not an exhaustive re-run of trial skirmishes.

By blending substantive IP clarification with a robust message on appellate discipline, the decision will influence contract drafting, litigation strategy and appeal advocacy in the technology sector for years to come.