For founders, executives, engineering leaders, and counsel

Who owns the code?

No human author. No copyright. Nothing to own.

If a person wrote it, you can own it. If an AI wrote it, no one can. Under current U.S. copyright law, code generated entirely by AI has no human author, so it cannot be copyrighted and cannot be defended as an asset you truly own, license, or sell.

Our guidance applies to code generated from every AI tool

  • ChatGPT
  • GitHub Copilot
  • Claude
  • Google Gemini
  • Microsoft Copilot
  • Cursor
  • OpenAI Codex
  • Perplexity
  • DeepSeek
  • Grok
  • Meta Llama
  • Mistral AI
  • Amazon Q Developer
  • Replit
  • v0 by Vercel
  • Bolt
  • Lovable
  • Windsurf
  • Sourcegraph Cody
  • JetBrains AI
  • Warp
  • Zed
  • Cline
  • Qodo

The four things teams miss

Four ways AI code puts ownership at risk

Ownership is not a formality you fix later. It is decided by who authored the work, line by line.

You cannot own what an AI wrote

Code written entirely by an AI tool has no human author, so under U.S. copyright law it cannot be copyrighted, and you cannot own it.

"Vibe coding" is real exposure

When you let the AI make the creative decisions, the code is not human-authored, and code that is not human-authored is not a protectable asset. Prompting, however detailed, is not authorship.

Mixed codebases are only partly yours

In a codebase that mixes human and AI work, you own the parts a person meaningfully wrote. The AI-generated parts stay unprotected until a human reworks them with real creative judgment.

Open source is not "no copyright"

Open source still has an owner; a license only works because someone holds the copyright. An open-source license placed on pure AI output has nothing behind it to enforce.

What the courts and the Copyright Office have said

The core rule is already settled

These are not predictions. Each is a recent decision or report that hardened the rule: no human author, no protection.

  1. March 2026 Supreme Court of the United States

    Supreme Court declines to hear Thaler; the human-authorship rule stands

    The Court declined to review the D.C. Circuit's ruling, leaving the human authorship requirement as settled U.S. copyright law unless Congress acts.

  2. March 2025 Thaler v. Perlmutter

    Works created entirely by AI are not eligible for copyright

    The court confirmed that only humans can hold copyright protection; a work generated wholly by a machine has no author the law will recognize.

  3. February 2025 Thomson Reuters v. Ross

    First decision to reject a fair-use defense for AI training

    The first ruling to reject a fair-use defense for using copyrighted material to train an AI, narrowing a shield many teams assumed they had.

  4. January 2025 U.S. Copyright Office report

    Only meaningful human creative input earns protection

    The Office stated that only the portions of a work carrying meaningful human creative input are eligible for protection. The AI-generated remainder is not.

It is all academic until you get sued, sell the company, or sign the contract

Find out what you actually own

A short assessment identifies your organization's AI-code risk persona, maps where AI-generated code sits in your product, and shows where your ownership quietly breaks down.