Software developers are copyright holders just like artists, musicians, and filmmakers. But code behaves differently under copyright law, and rules written for creative works often misfire when applied to software. GitHub has been pressing that point with the US Copyright Office (CO) during a series of consultations on "technical measures"—the upload filters and automated scanners that lawmakers and regulators have pushed online platforms to adopt. In its closing session, the CO agreed with a key GitHub concern: automated filtering does not work well with software, because determining infringement requires human review of content, context, and the consequences of removal, not automatic takedowns of matched code.

Where automated filtering goes wrong

Code defies simple detection

Source code on GitHub is generally open source and distributed without payment. Copyright protection extends only to the expressive parts of code, and identifying those parts is not a task that automation handles well. Code is also a moving target—rarely finished, regularly updated, and stored with version control systems like Git that make specific pieces hard to isolate and remove via filtering technologies.

Detection tools trip over code's context

Technical measures built to catch infringement in text, images, or video do not translate to source code. Similar or duplicate code often gets flagged in error, because detection tools do not understand the context that copyright law requires before a use is deemed infringing. Those false positives carry outsized risk in interdependent code ecosystems, where code is built to be shared and reused. GitHub already routes copyright takedown notices through technical review precisely because of this complexity.

Removal contradicts how open source works

Much of the valuable content on code-hosting platforms is licensed for sharing and remixing. Automatically pulling code down runs against those norms—when developers publish under an open source license, they generally want their code distributed. Disputes on GitHub are often not about whether code may be posted at all, but about meeting license terms such as attribution. Those conflicts are frequently resolved not by removing code, but by a developer adjusting the code or its accompanying license.

Takedowns ripple through dependencies

Code hosted on GitHub may be in use on millions of machines. A wrongful takedown can disrupt not only the repository owner but the broader developer ecosystem. Modern software leans on tested, widely available open source components written by third parties; mobile apps and enterprise systems alike depend on these packages. Removing a single dependency in response to an infringement claim can break an exponential number of programs that rely on it.

Pushing for developer-aware policy

GitHub has a track record of advocating for developer interests in copyright policy. It helped carve software development platforms out of upload filter requirements in the EU Copyright Directive, with developers speaking up about the potential impact. The company also runs its DMCA Takedown Policy with an eye toward explaining how the Digital Millennium Copyright Act applies to developers, and it funds a Developer Defense Fund that provides independent legal help through the Stanford Juelsgaard Clinic for developers facing DMCA cases. With the CO's consultation findings now reflecting the limits of automated filtering for software, GitHub intends to keep pushing for future copyright rules that account for the distinct norms of code collaboration. Recordings of all consultation plenary sessions are available from the Copyright Office.