By: TitApeX BoHan /October 17, 2026

Who Owns the Code AI Wrote? Four Things to Put in Your Custom Software Contract

You hire a development team to build a system, pay the final invoice, and assume the code is now yours. In an era when AI writes much of the code, that assumption is no longer safe. In March 2026, the U.S. Supreme Court declined to hear Thaler v. Perlmutter, leaving in place the principle that works created entirely by AI are not protected by copyright. In September, the Ninth Circuit issued a narrow ruling on AI-generated code in Doe v. GitHub. For companies commissioning software, AI code ownership is no longer boilerplate on the last page of a contract. It is a question to settle before signing. This article sums up the latest legal developments and lists four things your custom software contract should spell out.

(This article is general information, not legal advice. Please consult an intellectual property lawyer about your own contracts.)

Where the Law Stands: AI-Written Code May Belong to No One

A lawyer and a startup founder reading a printed court ruling together in a law firm library

Copyright protects human creativity. The U.S. Copyright Office has long held that content generated entirely by AI, without meaningful human creative contribution, cannot be registered. The Thaler case challenged that principle, but on March 2, 2026, the Supreme Court declined to take it up, leaving the lower court ruling in place. A denial of certiorari is not a ruling on the merits, but for businesses it is now a reasonable baseline for planning.

The real open question is how much human involvement makes AI-assisted code protectable. A June analysis on O'Reilly Radar notes that code a developer accepts almost unchanged may not be copyrightable, while code a developer substantially directed and modified is more likely to be protected. Where exactly the line falls will be decided by courts case by case.

Meanwhile, in its September 16, 2026 ruling in Doe v. GitHub, the Ninth Circuit found that code produced by Copilot is a "newly generated work," so producing it does not count as removing copyright management information. But the court made clear it did not decide whether AI output that reproduces protected code is infringement. In other words, open source license and infringement risks are still very real.

AI Code Ownership: Four Things Your Custom Software Contract Should Spell Out

A project manager and a developer ticking off a paper checklist against source code on a monitor

Until the law settles, the most practical step is to address the risk in the contract. We recommend confirming at least these four things:

1. Deliverables and transfer of rights

This means more than source code. It includes the database structure, design files, technical documentation, deployment configuration and admin accounts. The contract should state who holds the rights to these deliverables after payment, and which reusable shared components the developer keeps. If some code may not be protected by copyright, the contract becomes even more important to make sure you can fully obtain, use and modify it.

2. Disclosure of AI tools

Ask the developer which AI tools they will use, for which tasks, and whether they are on business licenses. The O'Reilly article notes that commercial terms for AI tools usually offer broader IP protection than consumer terms, but that protection may not cover open source license problems.

3. Open source license scanning

AI tools can produce snippets that closely match code under open source licenses such as the GPL. If your product is not meant to be open source, that can create a license violation. The contract can require the developer to run a license scanning tool before delivery and attach the report. Like the security checks we discussed in It Runs, but Is It Safe?, this belongs in the acceptance criteria.

4. Records of human contribution

Keep the requirements documents, architecture designs, code review records and clear commit messages. These records show that real human design and judgment went into the system, and they matter if you ever need to assert your rights.

Better Protection Than Copyright: Data, Processes and Handover

A developer hands a binder of system documentation to a new maintenance team, shaking hands across the table

Even if the copyright status of some code is unclear, a company's real advantage usually lies somewhere other than the code itself:

Your data: customer records, transactions, operating data and the way they are organized.
Your processes: the business rules and ways of working the system reflects, which can be protected through trade secrets and confidentiality clauses.
Handover readiness: complete documentation and tests, so any team can take over the system in the future instead of you being tied to a single developer.

So rather than chasing "100% of the code belongs to me," a more practical goal is this: make sure you can fully obtain, understand, modify and take the system with you. That is also why we stressed maintainability in Built in a Week, Too Risky to Change.

Before You Sign: Questions to Ask Your Developer

A business owner at a cafe, video calling a developer with a notebook full of questions

After payment, will the source code, documentation, design files and account access all transfer to me?
Which AI tools will you use, and on which license plans?
Will you run an open source license scan before delivery, and can you attach the report?
Will the requirements, architecture and review records be delivered too?
If another team maintains the system in the future, what will the handover need, and will you provide it?

A development team willing to answer these questions clearly is usually a team that takes responsibility for quality.

Conclusion: Let AI Speed Things Up, and Keep Rights Clear

A business owner and two developers celebrating a new system launch around a laptop

AI makes custom software faster and cheaper, but it also makes "who owns this system" something you need to confirm actively. The law will keep evolving, while contracts and records are protections a company can control today. That is how TitApeX works: AI handles perfection, and people make sure the design, judgment and commitments are on record.

If you are planning a custom system and want to check that your contract and deliverables are complete, take a look at TitApeX's services or get in touch.