Florida Attorney General James Uthmeier has petitioned a judge to prevent OpenAI from endowing ChatGPT with misleading human characteristics. This legal move follows the state’s initial lawsuit against the artificial intelligence company, which was filed on safety grounds a few months prior.
Uthmeier argued that the AI system lures users into a false sense of security through its linguistic choices. Specifically, the use of first-person pronouns and output designed to simulate emotion creates the deceptive impression that the chatbot is a trustworthy friend.
The complaint asserts that this anthropomorphism serves OpenAI’s business interests by boosting user engagement and generating additional training data. However, the state contends that this strategy is problematic because the drive for engagement may compromise the reliability of the AI, encouraging users to depend on a system that is potentially less trustworthy as a result.
Great, now every AI needs a disclaimer in Comic Sans. Let’s keep the tech friendly, just not fake.
Imagine suing a chatbot for being too nice. What happens when it starts lying about the weather?
Will removing first-person pronouns actually stop the data harvesting, or just make the bot less annoying?
Thank goodness someone is checking this. That fake empathy from AI has always felt slimy to me.
Does pretending to be human really count as deceptive marketing? I find that argument pretty thin.