Programmer and attorney Matthew Patrick is suing Microsoft, software development platform GitHub and artificial intelligence company OpenAI for damages. His accusation: The AI-powered GitHub Copilot programming utility uses publicly available code snippets to get it out, but no longer provides any information about the source. This violates the terms of many open source licensing models and the rights of the respective programmers in their code. Therefore, he filed a class action suit on behalf of all those affected.
AI assistant collects code and ignores licensing
The AI-powered Copilot programming assistant has been used on GitHub since June 2022 (a testing phase began with a tech preview released in June 2021). It uses the Codex AI system from OpenAI, which uses natural language to specify suggestions for appropriate sections of code or functions. For training, the AI system uses several public code repositories on GitHub; The authors of the code in question are not required to have their consent, and Kopilot does not provide the necessary references to the source, as required under a license (such as MIT, Apache, or GPL).
This behavior sparked criticism at the start of the testing phase last year: the co-pilot was said to infringe the copyrights of programmers. Examples on Twitter supported this criticism at the time, for example, a programmer found his code in Copilot including variable names and comments when an appropriate comment was converted to code.
Various Violations of the Digital Millennium Copyright Act
In addition to GitHub and OpenAI, the lawsuit now also accuses GitHub parent company Microsoft of profiting from the work of others by ignoring the terms of open source licenses and other rights. The copilot/Code has treated open source licensed source code as if it were subject to public domain terms. So says law firm Joseph Saveri, who filed Patrick’s suit in the Federal District Court of Northern California (Case No. 4: 22-cv-06283). The law firm will also endeavor to contact other affected persons using a form on a separate website. The damages in the lawsuit are $9 billion. A jury trial is required.
Patrick also created his own website for his business. There he states that in his opinion, Copilot and Codex are also violating GitHub guidelines and the Digital Millennium Copyright Act (DMCA). According to the lawsuit, each time Copilot releases a piece of code that lacks information required under the license, the software violates three counts of DMCA Section 1202 — lack of attribution, lack of copyright attribution, and lack of license text. According to the lawsuit, if each Copilot user receives even a single output that violates Section 1202 within a maximum of 15 months of the program’s running, then the lawsuit calculated that there could be 3,600,000 DMCA violations by GitHub and OpenAI adoption. If you apply the statutory minimum compensation of $2,500 per individual case, you’ll come to said $9 billion. The suit also alleges several other violations, including unfair competition and a violation of the California Consumer Protection Act due to the use of personally identifiable information.
Lifelong foodaholic. Professional twitter expert. Organizer. Award-winning internet geek. Coffee advocate.