GitHub Copilot is supposed to make daily tasks easier for developers by completing code using an AI. However, in some cases the program has copied copyrighted code. Microsoft will soon have to answer for it in court.
To combat possible copyright infringements, attorney Joseph Saveri and open source attorney and developer Matthew Butterick have filed a class action lawsuit against Microsoft, GitHub, OpenAI and other groups.
@github copilot, with “public code” blocked, broadcasts large chunks of my copyrighted code, no attribution, no LGPL license. For example, the simple prompt “sparse matrix transpose, cs_” produces my cs_transpose in CSarse. My code on the left, github on the right. Not okay. pic.twitter.com/sqpOThi8nf
—Tim Davis (@DocSparse) 16 October 2022
The artificial intelligence tool Copilot is based on a model from OpenAI and is said to have been trained using billions of lines of code found on the internet. However, the open sources used for this purpose were not necessarily approved for this purpose. In mid-October, for example, Tim Davis had noted that Copilot had issued proprietary code for transposing a sparse matrix.
Since this is a class action approved in the US, other affected developers also have the opportunity to join the case.
In the complaint filed, GitHub Copilot is accused of using code from Quake III. The fact that the software is under the GPLv2 license was not indicated after the content was copied. In principle, copyrighted source code may only be used under the conditions specified in the license. It is also illegal to remove author metadata.
In connection with the GitHub Copilot, the question of whether code produced using an artificial intelligence tool can be copyrighted may arise in the future. Currently, there is a difference between self-written software and software generated by another program.
That is why some jurists should still examine the issue in the coming years.