Last updated: 2026-11-18
Making More of Your Output Protectable
Copyright arrives automatically, but it doesn't arrive evenly. The parts of a project that reflect a programmer's own creative choices are protected; the parts dictated by a specification, a framework, a formula or a model's suggestion may not be. Who owns the result depends on contracts as much as on authorship, and proving either later depends on records made at the time.
This page sets out what programmers can do to make more of what they produce protectable, to own what they think they own, and to be able to show it. It covers where creative choices live in software work, evidence, ownership, AI-assisted code, clean-room discipline, and when to use something other than copyright.
Where Your Creative Choices Live FoundationalKnowledge that endures for decades โ core principles
The UK originality test asks whether the author made "free and creative choices so as to stamp the work created with their personal touch"1. Effort doesn't count on its own; choice does. Software work has more room for choice in some places than others:
| Artefact | Room for creative choice | Notes |
|---|---|---|
| Architecture, decomposition, module structure | High | Recorded in design documents, which are protected as "preparatory design material for a computer program"2 |
| Documentation, tutorials, comments | High | Prose is where originality is easiest to show; in SAS v World Programming the manual's wording was the part found to be copied3 |
| Screen designs, icons, visual layout | High | Can be artistic works in their own right, separate from the program4, and icons can be registered designs |
| Ordinary application code | Moderate | Naming, structure, sequencing and organisation all involve choice |
| Code that transcribes a formula or standard | Low | The constraints dictate the result |
| Generated boilerplate | Very low | Authorship lies in the specification and templates, not the output |
None of this means writing worse code to look creative. It means recognising that the design thinking, explanation and organisation around the code are often the most clearly protected parts of a project, and treating them as deliverables rather than throwaway notes.
Keep Evidence of Authorship and Date Applied / MethodologicalKnowledge with a 5โ10 year half-life โ stable practice
No registration or notice is needed: under the Berne Convention "the enjoyment and the exercise of these rights shall not be subject to any formality"5. The cost of that is that nothing official records who wrote what, or when. Since infringement requires copying and a file's bits don't show where they came from (see Why Maths Isn't Copyright), disputes are decided on evidence of origin.
- Use version control from the first line, with commits attributed to real authors and timestamps that a third party can trust, for example a hosted repository.
- Keep design documents and decision records dated and versioned alongside the code.
- Add copyright and licence notices to files and repositories. They aren't required for protection, but they tell others who owns the work and on what terms they may use it.
- Record what you brought in: third-party libraries, copied snippets, generated code and their licences.
Make Sure You Own What You Think You Own Applied / MethodologicalKnowledge with a 5โ10 year half-life โ stable practice
Authorship and ownership are different questions. Under the CDPA:
- Where a work "is made by an employee in the course of his employment, his employer is the first owner of any copyright in the work subject to any agreement to the contrary" (s.11(2)).
- A freelancer or contractor who writes code owns the copyright unless they assign it, and an assignment "is not effective unless it is in writing signed by or on behalf of the assignor" (s.90(3)).
- An exclusive licence must also be in writing and signed (s.92(1))2.
A business that commissions software without a written assignment often holds only an implied licence to use it, while the contractor keeps the copyright. Students should check their university's IP policy and the terms of any placement or project sponsor before assuming they own their work, and before giving it away.
AI-Assisted Code Ephemeral / ToolingKnowledge that evolves in months to a year โ check for updates
Code produced wholly by an AI model from a short prompt may not be protected at all. UK law currently treats a work with no human author as computer-generated under section 9(3) of the CDPA, but the government proposed in March 2026 that this protection "should be removed, while copyright should continue to protect works created with AI assistance"6. To keep your contribution protectable:
- Do the design yourself and record it: the structure, interfaces and decomposition the generated code fits into.
- Select, arrange, rewrite and integrate generated code instead of pasting it unchanged, and keep the history that shows you did.
- Note which parts were generated. That helps you assess both your own protection and the risk that generated code reproduces someone else's.
The position is developing quickly; see Is a Binary a Copyright Work? for the current law.
Clean-Room Discipline Applied / MethodologicalKnowledge with a 5โ10 year half-life โ stable practice
When you build something compatible with another product, protecting yourself matters as much as protecting your own work. A clean-room process separates the people who study the original from the people who write the new product. One group examines the original's documentation and behaviour, and writes a functional specification in neutral form: inputs and outputs, test cases, syntax diagrams. A second group, which has not seen the original's code or read its documentation closely, builds from that specification alone. Records of who saw what, and of everything passed between the groups, let you show later that resemblances come from shared function, not copying.
SAS v World Programming shows where the risk lies. WPL never saw SAS's source code, so its program was never in danger. Its manual was: the writers "had the language of the SAS Manuals firmly in their minds", and although "they tried to avoid describing the functions in the same language, they did not succeed"3. Apply the clean-room separation to documentation as strictly as to code.
When Copyright Isn't the Right Tool Applied / MethodologicalKnowledge with a 5โ10 year half-life โ stable practice
| If what you want to protect isโฆ | Consider | Act beforeโฆ |
|---|---|---|
| Unpublished source code, algorithms, model weights | Trade secret: keep it secret by "reasonable steps"7 such as access controls and confidentiality agreements | Anyone sees it without an obligation of confidence |
| A new technical solution implemented in software | Patent | Publishing, demonstrating or releasing it: anything "made available to the public" before filing counts against novelty8 |
| Icons, typefaces, the look of a screen | Registered design: "product" includes "graphic symbols" and "typographic type-faces", though not a computer program itself9 | 12 months after you first disclose the design yourself |
| A dataset you invested in collecting or verifying | Database right10 | It arises automatically; keep records of the investment |
| Your product's name or logo | Trade mark, for the goods and services you intend to use it on | Launching under the name |
Beyond Copyright covers each of these in more detail.
What You Can't Make Protectable FoundationalKnowledge that endures for decades โ core principles
- Functionality, programming languages and data formats. Others may reproduce what your software does, provided they don't copy how you expressed it.
- Ideas, algorithms and mathematical methods as such. Copyright protects your implementation; only a patent can cover a technical application of a method.
- Effort alone. Obfuscating, padding or complicating code adds no protection; what counts is the creative choice behind it.
Related Topics
- Software and Intellectual Property: A Map for Programmers: the hub for this set.
- SAS v World Programming: the case behind the clean-room advice.
- How University IP Contracts Close Every Exit: ownership clauses in an academic setting.
References
-
THJ Systems Ltd v Sheridan [2023] EWCA Civ 1354, para 16. https://caselaw.nationalarchives.gov.uk/ewca/civ/2023/1354 โฉ
-
Copyright, Designs and Patents Act 1988, ss.3(1), 11(2), 90(3) and 92(1). https://www.legislation.gov.uk/ukpga/1988/48/contents โฉโฉ
-
SAS Institute Inc v World Programming Ltd [2010] EWHC 1829 (Ch), paras 148 and 319. https://caselaw.nationalarchives.gov.uk/ewhc/ch/2010/1829 โฉโฉ
-
Berne Convention for the Protection of Literary and Artistic Works (Paris Act 1971), Article 5(2). https://www.wipo.int/wipolex/en/text/283698 โฉ
-
Department for Science, Innovation and Technology, Department for Culture, Media and Sport, and Intellectual Property Office. (2026, March 18). Report on Copyright and Artificial Intelligence. https://www.gov.uk/government/publications/report-and-impact-assessment-on-copyright-and-artificial-intelligence/report-on-copyright-and-artificial-intelligence โฉ
-
Trade Secrets (Enforcement, etc.) Regulations 2018 (SI 2018/597), reg 2. https://www.legislation.gov.uk/uksi/2018/597/regulation/2 โฉ
-
Patents Act 1977, s.2(1)โ(2). https://www.legislation.gov.uk/ukpga/1977/37/section/2 โฉ
-
Registered Designs Act 1949, ss.1(3) and 1B(6)(c). https://www.legislation.gov.uk/ukpga/Geo6/12-13-14/88/section/1 โฉ
-
Copyright and Rights in Databases Regulations 1997 (SI 1997/3032), reg 13. https://www.legislation.gov.uk/uksi/1997/3032/regulation/13 โฉ