TCAI Guide to ‘ChatGPT For Teens’: The good, the bad, and the law

At the Transparency Coalition we’re encouraged by the progress this represents: The maker of the world’s most popular AI chatbot is taking a real step to remedy the reckless and harmful roll-out of ChatGPT. We’ve also learned to be skeptical and realistic about the.

Many of the new safeguards embedded in ChatGPT For Kids are not purely voluntary measures implemented by OpenAI out of the company’s concern for kids’ safety. They are actually required by new state chatbot safety laws taking effect in Washington, Oregon, New York, Colorado.

What Happened

Those safeguards, which were strongly resisted by the tech industry and its lobbyists during the legislative process, are now embedded in the next generation of chatbots and presented as consumer-friendly features. With that in mind, we offer this early overview of ChatGPT.

• We strongly advise all users to refrain from entering any medical information into a chatbot prompt—and especially warn against connecting one’s personal Electronic Health Records into a chatbot.

• Is OpenAI planning to allow teens to upload their personal medical records in the future?

• TCAI comment: Again, this begs the question regarding the “yet” in that sentence.

Key Details

The teen product includes a number of designed safety features not included in the adult version. Some teens will want a ChatGPT product designed for them.

• It also raises questions about the use of the word “currently.” There is no pledge or guarantee that ads will not be shown to teens in the future.

• OpenAI currently shows ads to adults using ChatGPT’s Free tier or Go subscription.

• The state laws were prompted—and enacted—this past year in response to alarming and repeated instances of chatbots, including ChatGPT, encouraging teens to commit suicide, hide their self-harm and suicidal ideation from parents.

Why It Matters

Some will not; a certain percentage of them will want the full adult version. If OpenAI’s guardrails work correctly, they won’t get it.

• Those same state laws prohibit AI chatbots from exposing minors (under age 18) to sexual imagery or roleplay.

• The new state laws also require chatbot operators to identify user prompts and conversations that veer into self-harm or suicidal ideation, and respond by offering real-world support like a crisis center or other professional.

What Reports Say

Coverage of the story so far points to:

• Continued reporting by Transparency Coalition as more details emerge