That they engage in conversation in the manner of an assistant, a friend, a fictional character, or a knowledgeable stranger is a big part of why they're valued at billions of dollars. But the fact that chatbots say things - that they produce fresh claims, arguments, facts, or bullshit - is also a huge liability for the companies that operate them.
Chatbots perform roles associated with outside users - someone to talk to, someone with an answer to your question, someone to help with your work - but what they're doing is, in legal terms, much closer to automated, error-prone publishing.
"I don't think you get Section 230 immunity on the fly if you generate a statement that seems to be defamatory," says Mark Lemley, director of the Stanford Program in Law, Science & Technology.
Collection
[
|
...
]