Are your AI-based software platforms becoming a legal liability?
Image credit: Kiattisak Lamchan/Dreamstime
Successful prosecutions of companies developing artificial intelligence tools on the basis of copyright infringement could have significant repercussions for this rapidly evolving area of technology.
When ChatGPT launched last year, it took the tech world by storm due to the role it can play both as a smart coding assistant and, more generally, to help speed up the writing process for people in almost any walk of life. Since then, generative AI systems for a variety of uses have gained fans the world over. However, there are some potential legal issues on the horizon.
In the US, an AI-based software platform owned by Microsoft - GitHub Copilot - and OpenAI, which supplied the open-source code for training purposes, are facing a class-action lawsuit over alleged copyright infringement. The case was filed by programmer and lawyer Matthew Butterick and several anonymous members of the open-source community.
On the face of it, Microsoft, Github and OpenAI appear to have a robust defence, arguing that the complaints should be thrown out due to ‘lack of injury’ and ‘lack of an otherwise viable claim’. Importantly, Microsoft’s defence lawyers claim that no copying has actually taken place, as Copilot uses its own generative AI in the form of algorithms to advise software developers on coding by suggesting improvements and flagging potential issues. This is achieved through a process of aggregation, which suggests code based on the probability of what should come next.
According to the UK’s Copyrights, Designs and Patent Act 1988, anyone claiming copyright infringement, in the UK at least, is required to prove that copying has occurred of a ‘substantive work’. Even though open-source code is by its nature publicly available and software developers may be using it in good faith, with the intention of building platforms that generate yet more open-source code, licensing issues can arise if commercial entities are subsequently found to be using them to generate significant profits outside of the terms of the open-source copyright licence. The class action in the US will therefore set an important legal precedent in this regard, helping to determine whether or not developers acting in this way are infringing copyright.
While Github Copilot is the first generative AI system to attract a class-action lawsuit, many will be watching the case with interest. Platforms such as Microsoft’s ChatGPT, which was also trained by OpenAI’s Codex, and Google’s Bard AI chatbot, are both text-generating platforms that can be used by software developers and others to produce intelligent natural language responses at lightning speed in response to user questions. Other forms of generative AI include DALL-E, which generates images, MusicLM, which produces music and GET3D, which produces 3D objects.
Open-source licensing is a complex issue, with many different types of licences available, each with its own set of terms and conditions. Some licences allow for more freedom than others, and most software developers know that it is essential to understand the specific terms of each licence to avoid infringing upon the copyright of the original developer. Disputes in this area are relatively common.
In the case against OpenAI, the original developers allege that the company has violated the terms of the licences under which the open-source code was released. It is claimed that OpenAI used code from an open-source project without properly crediting the original developers or adhering to the terms of the licence. The validity of this claim will ultimately depend on the specifics of the licence in question and the actions of OpenAI.
Compounding the legal technicalities, there is a contradiction at the heart of this class action, which could undermine AI-powered innovation in the future. Open-source licensing is often used to encourage collaboration and innovation, and companies like OpenAI have been instrumental in advancing the field of artificial intelligence. If the company were to face prosecution on the grounds of copyright infringement for losses incurred due to a failure to meet the licensing requirements of the original code developers, this could have a chilling effect on innovation and collaboration involving the use of AI.
Overall, generative AI and GitHub Copilot are powerful tools that can help developers work more efficiently and effectively, potentially leading to the development of more innovative and advanced technology. A clear sign of its usefulness to the developer community is that we’re already asking ‘where would we be without it?’ Hopefully, we won’t have to find out.
Karl Barnfather is a partner and patent attorney at European intellectual property firm Withers & Rogers.
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.