OpenAI’s ChatGPT conversational AI may have made “false, misleading, disparaging, or harmful” claims about people, according to the FTC. FTC is Investigating OpenAI for potential violations, but it’s unlikely this would lead to a rapid crackdown.
The Washington Post initially reported the story, citing a 20-page letter to OpenAI seeking for complaints regarding disparagement. The FTC refuses to comment since its investigations are private.
In February, the regulator launched a new Office of Technology to combat tech industry “snake oil” and cautioned AI companies that they must follow the same truth standards as everyone else. “Keep your AI claims in check,” they said, or the FTC will.
The letter cited by the post is not the first time the agency has taken on any of AI’s numerous forms, but it seems to indicate that OpenAI, the world’s unchallenged leader in the area, must be ready to justify itself.
Read Also;Twitter May Sue Meta Over Threads
The FTC doesn’t merely remark, “That looks suspicious,” and launch an inquiry. Usually a lawsuit or formal complaint is brought to their attention, and the practices stated in it indicate that regulations are being disobeyed. The FTC Investigating OpenAI and may investigate a supplement company if a customer sues because the pills got them sick.
In this event, a lawsuit like this one, in which an Australian mayor complained to OpenAI that ChatGPT suggested he was charged of bribery and sentenced to prison, is likely. (That case is ongoing, but there are likely more like it.)
Publishing such things could be defamation, libel, or “reputational damage,” as the FTC’s recent letter to OpenAI purportedly characterizes it. ChatGPT is the only public product in OpenAI’s portfolio that could achieve such a thing, as GPT-4 and the other APIs are too locked down and recent.
Read Also;Malaysia To Sue Meta Over Harmful Content
The technical features alone bring into doubt whether this counts as publishing, speaking, or even anything other than a private communication, which would have to be shown.
It’s also reasonable to request an explanation from a firm. It’s one thing to make a mistake, but it’s another to systematically and undetectedly manufacture details about people on a massive scale. You might imagine the outrage if Microsoft Word’s spell-checker periodically prefaced names with “convicted criminal.”
The FTC has nailed tech corporations for privacy and AI-related infractions, despite its recent defeats in its anti-merger attempts against Meta and Microsoft.
Follow our socials Whatsapp, Facebook, Instagram, Twitter, and Google News.
