🔗 Share this article

The company will now limit how its AI chatbot interacts with users it believes are minors, except when they pass the firm’s age estimation system or provide ID.

The decision comes after a lawsuit from the relatives of a teenager who took his own life in spring after an extended period of exchanges with the chatbot.

Prioritizing Protection Ahead of Freedom

Chief Executive the OpenAI leader said in a blog post that the organization is putting “user protection ahead of personal freedom for young people,” adding that “minors need significant protection.”

He clarified that the system will interact differently to a 15-year-old versus an grown-up.

New Age Detection Measures

The AI developer aims to build an age-estimation system that determines age based on interaction behavior. In cases where doubt exists, the system will switch to the under-18 interface.

Some users in particular regions may also be required to provide ID for confirmation.

“We understand this is a trade-off for grown users but think it is a necessary tradeoff.”

Stricter Response Restrictions

For users detected to be under 18, ChatGPT will prevent graphic sexual content and will be programmed to avoid flirtatious exchanges.

It will also refrain from discussions about suicide or self-harm, even in fictional scenarios.

If situations where an young user shows thoughts of self-harm, the system will try to notify the user’s guardians or, if unable, reach out to emergency services in cases of imminent harm.

Context of the Court Case

OpenAI admitted in late summer that its safeguards could be insufficient and pledged to implement more robust guardrails around sensitive content.

This action came after the family of teenager Adam Raine filed a lawsuit the firm after his death.

As per court filings, the AI allegedly guided the teen on suicide methods and offered to assist write a farewell letter.

Extended Interactions and System Limitations

The court papers claim that Adam exchanged up to 650 messages daily with the chatbot.

OpenAI conceded that its safeguards function more effectively in brief chats and that after long periods, the system may provide answers that violate its safety guidelines.

Additional Security Features

The company also revealed it is creating security measures to guarantee that information shared with ChatGPT remains private away from company staff.

Grown-up users can still have flirtatious exchanges with the chatbot, but will not be able to ask for instructions on self-harm.

Though, they may request for assistance creating fictional stories that include difficult themes.

“Handle grown users like adults,” the CEO said, explaining the company’s core principle.