Why This Matters Now
Picture this:
A two‑person startup in San Jose is rushing to ship its MVP. They’ve pulled in a few popular GitHub libraries, let GitHub Copilot autocomplete big chunks of their backend, and pushed everything to production. The product works. The demo impresses investors. No one has really stopped to ask: “What licenses are attached to all this code?”
At the same time, across the Bay, one of the developers whose code helped build the modern open‑source ecosystem is starting to notice something strange. Code that looks very close to her carefully crafted functions is popping up as Copilot suggestions for other people—without her name, without the original license text, and without any of the conditions she chose when she released it.
These two realities collided in a real case in the Northern District of California: a lawsuit against GitHub, Microsoft, and OpenAI over how GitHub Copilot and related AI tools use open‑source code and handle (or ignore) the licenses attached to it.
In 2024, the federal court made a critical move: it threw out some of the pure “IP” claims, but allowed key breach‑of‑license and contract claims to go forward. In other words, the court signaled that open‑source licenses are not just suggestions—they are enforceable agreements.
For developers and startups in California, this matters because:
- You may be relying on tools and workflows (like Copilot, ChatGPT, or code from GitHub) that quietly bring in licensed code.
- You are still legally responsible for the code you ship, even if an AI assistant wrote part of it.
- Treating “open‑source” as “free for anything, with no rules” is no longer just a bad engineering practice; it can become a real legal problem for your product and your company.
At Yang Law Offices, we regularly work with entrepreneurs and professionals whose IP is central to their business—whether that’s code, content, or branding. Understanding what this kind of case means doesn’t require a law degree. It just requires knowing that licenses are promises, and California courts are starting to enforce those promises in the age of AI‑assisted development.
What This Case Is About
The lawsuit often called the “GitHub Copilot case” is officially a group of developers suing GitHub, Microsoft, and OpenAI in federal court in the Northern District of California. It focuses on how GitHub’s AI coding assistant, Copilot, and the underlying models (like Codex) use open‑source code hosted on GitHub.
Here’s the core of it, without legal jargon:
Who Sued Whom?
A group of programmers who publish code on GitHub under various open‑source licenses filed a class‑action style case against:
- GitHub – which hosts their repositories and runs Copilot;
- Microsoft – GitHub’s parent company and a major Copilot provider;
- OpenAI – which helped develop the AI models that power Copilot.
These developers aren’t saying “no one can ever use our code.” In fact, they chose open‑source licenses so others could use it—under specific conditions.
What Are the Developers Complaining About?
The developers claim that GitHub, Microsoft, and OpenAI:
- Took code from public GitHub repositories that was released under open‑source licenses (like MIT, GPL, Apache, etc.),
- Used that code to train AI models (Copilot/Codex), and
- Allowed Copilot to emit snippets that:
- Are identical or very close to their original code in some cases, and
- Do not include the required attribution or license notices that those open‑source licenses demand.
In other words, their argument is:
“We gave the world permission to use our code as long as certain license terms were followed. Your AI used our code and then reproduced it for others without following those terms.”
What Has the Court Done So Far?
The judge has not decided who is right yet—but there’s one important early ruling:
- The court dismissed some of the pure IP claims, such as certain DMCA (Digital Millennium Copyright Act) theories that required showing Copilot output was identical to the original code.
- But the court allowed the contract and license‑breach claims to go forward.
In practical terms, that means:
- The judge is willing to treat open‑source licenses as binding agreements.
- If a company (or its tools) uses licensed code in ways that ignore those conditions—like stripping attribution or failing to include the license—they can potentially be sued for breach of license/contract, even if the strict copyright or DMCA claims are harder to prove at this stage.
For developers and startups, this is the key takeaway:
Open‑source licenses are not “public domain.” They are permission slips with rules. If you build products—especially with AI‑assisted tools—you need to understand and respect those rules, because California courts are starting to enforce them.
Key Licensing Lessons from the Copilot Case
Open‑Source Licenses Are Contracts, Not “Free for All”
When you put code on GitHub under an open‑source license, you’re not giving up ownership. You’re giving other people permission to use your code under certain conditions.
Common licenses (MIT, Apache, GPL, etc.) are essentially contracts:
- “You may use, copy, and modify this code…
- If you follow these rules (like giving attribution, keeping the license text, or sharing modifications).”
In the Copilot lawsuit, the court let the developers’ breach‑of‑license / contract claims move forward. That means the judge is treating these open‑source licenses as real, enforceable agreements, not just nice‑to‑have credits.
For developers and startups, the message is simple:
If you accept the benefits of open‑source code, you’re also accepting the terms that come with it.
“Identical Copy” Matters Less for Contracts Than You Think
Some of the strict IP claims in the case—especially under the DMCA—required the developers to show that Copilot’s output was identical or nearly identical to their original code before they could even get to the next step. That’s a high bar.
But for license and contract claims, the question is different:
- Did you (or your tool) use licensed code in a way that ignored the conditions of the license?
If an open‑source license says:
- “Keep this notice,” or
- “Include this license text,” or
- “If you combine this code into your project, you must open‑source your project too,”
you can be in breach even if the AI assistant slightly reformats the code or changes variable names. What matters is whether the license obligations were honored.
Open‑Source = “Permission With Rules,” Not “No Rules”
A lot of developers casually treat “open‑source” as “public domain.” Legally, they’re not the same:
- Public domain: No one owns it; there are no rights to enforce.
- Open‑source: The author owns it, but gives you conditional permission to use it.
Some licenses are pretty light‑touch:
- MIT/BSD/Apache usually just want you to:
- Keep the copyright notice, and
- Include the license text and attribution.
Others are stronger:
- GPL/AGPL and similar “copyleft” licenses might require you to:
- Release your own combined or modified code under the same open‑source terms if you distribute it.
The Copilot case is a reminder that those conditions are not optional.
What This Means for Developers and Startups
For California devs and founders, the Copilot case boils down to this:
- You’re responsible for your stack.
It doesn’t matter whether a human, a library, or an AI assistant produced the code—if it ships in your product, your company is accountable for complying with any license attached to it. - You need basic license awareness in your workflow.
At a minimum, your team should know the difference between permissive licenses (like MIT/Apache) and copyleft licenses (like GPL/AGPL), and when attribution or sharing obligations are triggered. - AI tools don’t replace compliance.
Copilot, ChatGPT, and similar tools can speed up development, but they don’t “wash” licenses off the code they were trained on. You still need policies for reviewing, attributing, and documenting what goes into production. - Early cleanup is cheaper than late litigation.
A simple internal review of your dependencies, generated code, and licensing practices now is far less costly than dealing with a contract‑breach or IP dispute after your product is live and funded.
Who Needs to Know This
This isn’t just an issue for giant tech companies. The Copilot lawsuit has practical implications for:
- Solo developers and small dev shops who rely heavily on GitHub, open‑source code, and AI coding assistants.
- Early‑stage startups racing to ship an MVP, especially in SaaS, fintech, and health‑tech, where code is a core asset.
- Technical founders who write or oversee code but haven’t formalized any open‑source or AI‑use policies.
- Engineering managers and CTOs at growing companies who need to balance speed with legal and reputational risk.
- Product owners and non‑technical founders whose businesses depend on software, even if they don’t write a single line of code.
If your company’s value is tied to software in any way, understanding that “open‑source license = contract” should be on your radar.
Let’s Talk—When You’re Ready
If you’re a developer or founder in California, you don’t need to become a licensing expert overnight. But you do need to know enough to recognize when you might be taking on risk—especially as AI tools become part of everyday development.
Yang Law Offices is a Southern California‑based firm led by Attorney Elizabeth Yang, who brings both an electrical engineering background and deep IP and business‑law experience to her work. Our firm regularly helps:
- Developers and technical founders understand how licenses affect their codebase and products,
- Startups set up practical, lightweight open‑source and AI‑use policies, and
- Business owners protect the value of their technology when raising money, selling the company, or planning for the future.
If you’re unsure whether your current use of open‑source and AI tools lines up with the licenses involved, or you just want to make sure you’re building on solid ground, we’re here to walk through your situation in plain English and help you plan next steps at your pace.
Disclaimer
This blog is for general informational purposes only and does not constitute legal advice. Reading it does not create an attorney–client relationship with Yang Law Offices or any of its attorneys. Laws and court decisions can change, and how they apply depends on your specific facts. For advice about your situation, please consult a licensed California attorney.
Sources
- Joseph Saveri Law Firm – “GitHub and Copilot Intellectual Property Litigation”
Overview of the class action against GitHub, Microsoft, and OpenAI, alleging that Copilot violates open‑source licenses and documenting key case updates.
https://www.saverilawfirm.com/case/github-copilot-intellectual-property-litigation saverilawfirm.com - Butterick & Joseph Saveri Law Firm – “GitHub Copilot litigation” (case site and updates)
Background from plaintiff-side counsel explaining the theory that training Copilot on public GitHub repos violates the conditions of 11 popular open‑source licenses and GitHub’s own terms.
https://githubcopilotlitigation.com/ githubcopilotlitigation.com - The Verge – “The developers suing over GitHub Copilot got dealt a major blow in court” (July 9, 2024)
News report summarizing Judge Jon Tigar’s 2024 order that dismissed nearly all IP/DMCA claims but left open‑source license violation and breach‑of‑contract claims standing in the Northern District of California.
https://www.theverge.com/2024/7/9/24194596/github-copilot-microsoft-openai-lawsuit-dmca-open-source-licenses theverge.com - Joseph Saveri Law Firm – “GitHub Copilot Intellectual Property Litigation” (case updates section)
Confirms the November 3, 2022 filing date, the May 2023 motion‑to‑dismiss ruling, and the October 7, 2024 Ninth Circuit petition on the DMCA §1202(b) “identicality” question.
https://www.saverilawfirm.com/case/github-copilot-intellectual-property-litigation#case-updates saverilawfirm.com - Law Offices of Kate Downing – “An Open Source Lawyer’s View on the Copilot Class Action Lawsuit”
Practitioner analysis of the Copilot complaint, explaining its focus on DMCA §1202 and breach of open‑source licenses (rather than straight copyright infringement) and discussing how GitHub’s Terms of Service interact with OSS licenses.
https://katedowninglaw.com/an-open-source-lawyers-view-on-the-copilot-class-action-lawsuit/ katedowninglaw.com - TechTarget – “Developers warned: GitHub Copilot code may be licensed”
Explores practical risks for developers using Copilot, including potential license conflicts, and notes that the litigation centers on alleged violations of open‑source conditions rather than code being “public domain.”
https://www.techtarget.com/searchsoftwarequality/news/252527271/Developers-warned-GitHub-Copilot-code-may-be-licensed techtarget.com - GitHub – “GitHub Copilot · Your AI pair programmer” (product and responsible AI information)
Describes how Copilot is trained on public code, how suggestions are generated, and GitHub’s position on copyright, code matching, filters, and user responsibility for suggestions.
https://github.com/features/copilot github.com - GitHub – “GitHub Copilot · Plans & pricing” (FAQ sections on IP and responsible AI)
Additional details on Copilot’s training data, duplication filters for matching public code, and GitHub’s guidance on copyright and open‑source license compliance for users.
https://github.com/pricing/copilot github.com