The Electronic Frontier Foundation (EFF) and the Center for Democracy and Technology (CDT) have filed an amicus brief in Garcia v. Character Technologies. The case revolves around whether the output of a chatbot is protected by the First Amendment. So far, the case remains unresolved, but the judge has expressed skepticism that chatbot outputs constitute “speech” at all.
The lawsuit alleges that a teenage boy’s death was caused by interactions with a chatbot from Character Technologies.
The brief explains that the chatbot output is not the speed of a robot but reflects the expressive choices of humans. The choices that the EFF are referring to are the choices made by developers during reinforcement learning, where they give positive or negative feedback on responses. The choices of the user, such as the crafting of a system prompt, are also reflected in the output.
To be very clear, the EFF is not saying the chatbots have rights, it’s saying that human expression is being channeled through the technology, and that this is, or should be, protected by the First Amendment. It also argued that the user’s right to receive information is a protected right.
The EFF warns that if chatbot outputs have no First Amendment protection, then the US government could potentially ban chatbots that express viewpoints it disagrees with. The brief acknowledges that harms from chatbots can be regulated, but that such regulations must be appropriately tailored to avoid unduly burdening the right to express oneself.
The EFF is hoping that the brief can help the court understand how the technology works from a legal perspective. It is not a call for chatbots to be exempt from all regulation, but to ensure that regulations are constitutionally sound. Interventions like this from the EFF are really important right now, while AI is a nascent technology, to ensure that the law takes a balanced and fair view of issues related to AI.